diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/lm_head/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/lm_head/_0.pt
index 257f78e580e29f2b211849ea7bc8ce9dac122cc9..b26634d31627d1a0044c3f3651ff144ed5eb8f4c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/lm_head/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/lm_head/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/lm_head/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/lm_head/_s.pt
index eb091894ca1a6a000077745429b4b21f688defc3..8a23de69a32e79b5a1c8d012ae9ef99f818198cd 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/lm_head/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/lm_head/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.fc1/_s.pt
index 9cbf296d53646a3a2921f65a92a303dcdad352dc..c87a825dc0c1178aa377fa90be4519856f3ef5fe 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.fc2/_s.pt
index 6b642179592c2f2cb43d0e5b838cd31cabc0f402..774d39fcdefd7fcdc3a34f35b837f5a3b1aff972 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.k_proj/_s.pt
index 46b51da70f1c9fd541a6ca4a4d91ee753ecf3460..edd2329cecdb3d42b31a0f24fd8c76bd378e1d54 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.out_proj/_s.pt
index ba0238b187f1c3462c966a7fe90e9e9dec565b02..7321492ba279958a9986c00268c1c931187aebfe 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.q_proj/_s.pt
index e8fa2df6a81bbd392e1da5c76b13ea2990e357db..537d4cae3bf9829bcc3d236bb14d91e9aac3901c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.v_proj/_s.pt
index c00033ed1bc9d39701c3855809da9848dfcf7d36..c0a542cd09acb461c4f1cc4cebf03bea2beb9c47 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.0.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.fc1/_s.pt
index 6923bcbc368fda98afee7434dd552a7e78e56c65..8ea314ebe637e30e38e5e7746e285930102a885a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.fc2/_s.pt
index bb694159d8644354bbad3fec8c37a15e9a0d4755..ca6565a407d54624b88e29004b5dc6b8d7a14e92 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.k_proj/_s.pt
index b7f9842ab7ed64dfb1966d57e32eb7851d594029..f925e1c8bd28d5fc21edbf7f9bf254d45552f4a6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.out_proj/_s.pt
index 3266c13bec833d3cd639e54e575d1c39d91d97fe..f142cc931d9cff3fc3ad1621fee7ef686b1e18e4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.q_proj/_s.pt
index d357b70d3e4ced32fc79dc78c6c3f86827743b60..74798534acd7734b262e7c1144604f69d91a2bb4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.v_proj/_s.pt
index cced9a534ee333fc548636432df2baa3c8ff8845..963e838d1a3830c12c8e3719c82c7d61303f7d4b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.1.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.fc1/_s.pt
index 9bd91123a2ee85fa73d53d3813eff9f6bdcd8ba7..8caa3e4f0d002ca62d4552dc787e4be1a4b48659 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.fc2/_s.pt
index 177bfe27a93fdad63f51fcdadd04bc7b22eea878..8577e61904a0e1dd66e7edbea93845d38eda195b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.k_proj/_s.pt
index c10aa2a4525f75d2b48c2240cf0a340707cdd1dd..a47f53f359efcb2a41700997885f17096328daa8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.out_proj/_s.pt
index 7fe72b932deee013077948abdbe755594b58186f..e555c93c83d18989ec2e5ad327e02cf7b2353833 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.q_proj/_s.pt
index d2122e20cee44b680c5ecf93a087f609e5ffbd28..cd903a8db80194c3d2388f1a7b4d8154344c9e6b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.v_proj/_s.pt
index f249a5eb2d0d3bf15c2f4dd8b6ace505ba5d8d75..6adbc05f4c19c4cb157222df4ee8870f98061173 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.10.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.fc1/_s.pt
index aa749594cfb6857f5af59bcdbdd33e1c21fbea3c..d277ebc0c0b6e07b33ed7521c03cc1a70a9e402d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.fc2/_s.pt
index 4960d4155ff31e156f837b87e1e3661692d46683..ab921fc822da9592a0cc9a481a8cef0630989585 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.k_proj/_s.pt
index 9bf8e6875af61e127fa5d40409b63bf2ba696426..79528ed65774d64b5472573358a49d39e2c8aaca 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.out_proj/_s.pt
index 5073fd5f615fe8cccf86d83964f5d43f99ca9a34..242d15f77922cf6e14b4d3f50247d0fc9ed004a1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.q_proj/_s.pt
index 554a149908a424eef4906ee4570180d3e9af67fc..78f987bc66748f6ab435b0b96cad1e34479fd0e5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.v_proj/_s.pt
index 5d4d466851d079720bae29e4f231c09e284a999b..6e53b33851e828482188337d9556e7e18e1862e9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.11.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.fc1/_s.pt
index 6b507c47f717e924c89f67293b536f888111cebd..101e6ed736cd617487bc9bb30225f09d5dabf780 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.fc2/_s.pt
index cc02d8d6cffb38bcef184c4071eadad31fd683f7..151a139cfc0a852e496f2cb3f8c2d3a834927882 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.k_proj/_s.pt
index 7e040c3f1c862ad9f923309621986480c1572ae0..456a90589c4508a28f498e844e5327ae00250d8e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.out_proj/_s.pt
index a70f0e5668e5a52514c6ad4be6a9e3b27effc674..8d2840504550d5c76b4ee499a83728d0444fd6d8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.q_proj/_s.pt
index 7253ec3dcf452708c42ce98513160734f9df9ac7..e799d97bbbaeb074cd83334e9900a5ee8955ae2b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.v_proj/_s.pt
index b8ba162638347fcadb9000d621028ec60a11ad35..e75d6700f202d79a169c2346cc712a27d61bf4cd 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.12.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.fc1/_s.pt
index 123506f334ceec4b54015428b0433ffb5fd427af..262ba1584e0c3805b8a134f182bf2c13c32a721c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.fc2/_s.pt
index 41b81aced60bc8ce601f0d19a3426b3d7bf32ea1..2c3ba202f30d9671d1ebfd5851832a909ea8059c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.k_proj/_s.pt
index c3fe7a112b104b0c6e04b5cac9525b668a940a4c..2e5ef81df1fb69c00e128c90eff67833dfb7955b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.out_proj/_s.pt
index 752cd9c701c133c9a4ab34ccab2fe385e9b66180..8a969536c068fdab5a710acc644d140bd8b3bd44 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.q_proj/_s.pt
index ef6d50483a7fa0cbfd0e45265f3fda38e67b2f7a..72be07eddb25db95eba57196098dd3e2a9abc2cd 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.v_proj/_s.pt
index 748009735f06a4551f3acc79ccfa4ed14dabf9bc..cb1522683454867ae1f75b39bce7081a7c5df6e7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.13.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.fc1/_s.pt
index eddf2c1016399aeef9182de038ae7a67bdb7918c..d87a7eec5abd22bc250641be7c24556a33003e51 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.fc2/_s.pt
index 9ae04525add9c989d8d7ff6ca4d99296cdd83109..4f795eeabc20ee900a19568bb83ff46fc6e3b839 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.k_proj/_s.pt
index 7448d2757dc7b576e8c30f25cbc8c1043c2d5ecb..962cc4b8e100e519aeffc675a0c75fbbab912871 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.out_proj/_s.pt
index 9aa77fb032cd984ea5fc14132d7df331c1fcbac5..a4616f53f62df3022be6d04b6f0f3d962aad2122 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.q_proj/_s.pt
index 9ca96951b46bc5eb4dd256a13c54e35fcb7a590b..678b456ef7e823d0fa478a98ff806520855ef843 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.v_proj/_s.pt
index a21429a6a13f3e8892227a29d72084d944bff5c7..1e627a263450a2e6a1af673215bdd345d8bd01e5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.14.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.fc1/_s.pt
index 0ab4389f29b8efaea5851b5215a9385001a2747c..64f432a1c86f3ae2ac9878c4476d64a3b940a458 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.fc2/_s.pt
index c1310f4ee3851776f3615fb2ea73519daf7743b1..157c2268b177ca355c20d5d6259edd6f28d4cb30 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.k_proj/_s.pt
index dfbf07a4f0d0585c6ce1aa0da6568b2a50248f2f..710d55e421507d6b27446b8477fbae4f5e031148 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.out_proj/_s.pt
index 0d5fb36dfa4ef5517f247ae055a9b815c9f68c03..366afaa520c24fd5b9d45963384463580b001e8b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.q_proj/_s.pt
index 0d29deaf3efd55fb49dcad92c7ec27359c2e5972..4d61702ba884be2d294a2f810bd8d33d5374310c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.v_proj/_s.pt
index c319c89274e9de8d31d302db844e2d5ef0df43be..480679a832b9062970ddf1d8f0098539266dae71 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.15.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.fc1/_s.pt
index 67eafc0affff77e6c3319e4111a26ea7004b3220..352472a9e9a5a9bd958f4fb0919f96791e23d337 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.fc2/_s.pt
index 2323fd4037750ff1b81a77c53c7b92cc3a172dc0..a141038b523e517d14472f63547155ce3c552d7b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.k_proj/_s.pt
index aa3c5a95eb062b8c0d26e6c0ed7e751192bf6db4..14e79e850556861f5f0912f49f3a34c5a94a9305 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.out_proj/_s.pt
index ef1d383fc18f9185a597bf679f684e77df1bdb46..7572a9e99303044a2455f5bec02ee10f9439d47c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.q_proj/_s.pt
index 63a8a6bb553f55b8dfe9f7105a9d78540f7b54ca..d25116b725b831af6e7eda689d3d265c408d572f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.v_proj/_s.pt
index 7eb46d1e23f8bcc0056503ceda586d056dd31b45..3252df7fa47cfc77c6ee941938bca4e1a44a0baf 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.16.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.fc1/_s.pt
index 5853add0a2779d196f7ae2cfdc9bf624147f559a..731f4dee4fbc76a774164d6ef4e81b89f19e65d7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.fc2/_s.pt
index e90ffbff5ce2f1fc86ceeeea8113f9cfd64b46f8..0a961fd3d82131b257a0fe27b6790e344a14bec3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.k_proj/_s.pt
index da8c5d1f44f75c47ea8e10037ee8f85382bd5c68..e4f3515c2ce1c3c452b407f0e629737cf5f6e0f2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.out_proj/_s.pt
index f06cde449b3bb0272f060faf62d125c5a173656f..a6a129b6e10e18b00a2a5de68ed2472d979a9125 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.q_proj/_s.pt
index 0e58e2f5c0de543e63d09d5f7b25aa309d91e52d..a89776bac35be7c4f921c004e777089f191cc7ed 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.v_proj/_s.pt
index cf9f1b86817e758e8d4432df3ddae925c78d9048..71080b7571a19b14c2429eeee62bb4d70a81d81c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.17.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.fc1/_s.pt
index 8b050c534da0dd4159e944563b420ec53a6ce93d..ebdfd5f47338eed74d2a4f36d17567b367ad166e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.fc2/_s.pt
index ba976164b97753040fcd163f7a9b1ec76e25853d..5a89c09b125608b421d79c0a4dacadd7fffafafd 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.k_proj/_s.pt
index 47b89a9218468bcfc19ca83ad9267fb143d38f8c..1a6aea7cb7c36b9b2c7e11dead08a5fac73652ba 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.out_proj/_s.pt
index 1709535bb43d4a549b14620e9877ab80bb6fa14c..cef326c25a4b788a235cfe7b5bea43296c24de4b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.q_proj/_s.pt
index a6be61303212bd50ccb775070ce9be925ec7b59c..0826e5274ac28f3da687f376a2d43d6c83bc737e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.v_proj/_s.pt
index 2b4fed50beed079e76fd942a99f8f1d70ce72a1f..d4245348ef5a78cb018da9ba1e051e4510a51bbd 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.18.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.fc1/_s.pt
index a6555794ed575baf3199646746b1ba00973bf530..689ee87aca045d5a4af3e1fe22ef35f29e70c78c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.fc2/_s.pt
index f5933e52d0b0547d2bcef4dfa24381d7f293c6bf..c207a18e994ce1971fd5bcdfa3125d484a61f77c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.k_proj/_s.pt
index 28ca18877356b8c482be5b13272ecc3deebad06e..d9a0af0f63007b2ec3c6a29d6140e420d802a1d1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.out_proj/_s.pt
index 62448d6674f795f7fe95909b5462a2c2b9b12cc9..72645851d412a44d23e4ae94a4d5bf9f3944821b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.q_proj/_s.pt
index 3b887be500c9c85cf656d4b74c912efb22e066f1..2f25437adb0a273b9a148198691a79946efdc1d6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.v_proj/_s.pt
index cf41484ab4c4254ee22da1840124902ec7f6f5ce..b4e10dcd00f3c7f938b036910164dfcae7624a12 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.19.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.fc1/_s.pt
index 57b5c9f9162e9590b7760b3724ea42cb0c5f8c42..82d547d4f2a63d19af3fb32e90fd74182d17d9e3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.fc2/_s.pt
index e7130af83c180a8d5f145a1c42211a2bd2c3b2fa..271556462f66cb1cf8640e252be3e8690edd3109 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.k_proj/_s.pt
index 6ddb036b194e63fea537e8dfa6d60185b8588155..ee578a339aec407324d2644162ef255934411588 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.out_proj/_s.pt
index 34190956600ce1a65822d4d77c646d81915ab445..75a9d3312d323c1abfa66e4208903477d0608a09 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.q_proj/_s.pt
index d3f412964e73649a8f73f613bcb177dbf86e149f..6332d26d907432e2abad9f7e38326f1ca2c633e5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.v_proj/_s.pt
index 1b1678e2f54627e9199e1647720cc95c639f1637..4a4d00c5c2eb82b9ba71e66700956b7215f6562c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.2.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.fc1/_s.pt
index d939801b641ec137bb9f0a6c83a0046df62521be..cd8b0feedf79e95ecf62b75ea60a07dcb821f62a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.fc2/_s.pt
index e8b13ef6282628526ee08baeb3d01ef3237c5929..5543ff3358450285fe58e2b46fc06f557b4e90e9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.k_proj/_s.pt
index f21f7363b789e02b5a474ac28f634d57563602d6..62bca84b5f2b9966c1590789449a2fb5eb549fd7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.out_proj/_s.pt
index c8d09471f890ea1249aa4f3296a64bd20256b8d3..a75a55b24638a29eb0c628ee089f05de1e9d50a2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.q_proj/_s.pt
index 12eca964c5237453dabb4b5ce804dde814d35b4a..cdf1c4bcf60dc05ae7e385bac8a9a0fa18df6ba2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.v_proj/_s.pt
index 5bc105be9f9edf452ebe05589b31a7026995b1bb..3e442a0828521cf7d0b5f92e8709f6118291803b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.20.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.fc1/_s.pt
index bd02f3b498c527f1577b11d337a591483f95a35a..11e17132d953d879f6b8108623e06ec1bd59fd8e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.fc2/_s.pt
index 98960f550d96653a3258621ce3cf25a0f91e6c36..176ceb032f5d1a9824276b3d840b625924da39fd 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.k_proj/_s.pt
index e64775aca20c79d7c1cbbc0b9bf62050642d3315..71d14f5bbcf120a647a038d23d87ca41d08e9ea8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.out_proj/_s.pt
index d2ac8e29813e1edcb16338b0c8660237ace1431d..df37b32bbbd3c6dad507525656e9dcfe7d4c8281 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.q_proj/_s.pt
index 65bc4f7816a1fcf7bce5ff3c4b6a96663d38bcac..3e1b1e762bc91d2648b8c22217db8cc0be009288 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.v_proj/_s.pt
index 51f500dcea44f3263ab1180c25d79dcd99356e97..8a526f77b7eac7fc96b7ba72559f1cb95c7ac27c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.21.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.fc1/_s.pt
index 21a095a426054a976af5b15d1ec3375fe53a123d..0f9052b3f460f6b569d077b334cadc597112fa0f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.fc2/_s.pt
index 64afa74fd82c6751cd249d4868674280028c5ca0..eb9617f66ffb7dc2048600cb6bca39b83b27fd39 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.k_proj/_s.pt
index b08497f9f4bac16f4a76f96db4da9ec2cca21f4f..51282ba09d3aa73e578d1e0ba299714866c004ea 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.out_proj/_s.pt
index b5a3c4812e7e328cd8304e385a7caa5887a3e007..8fb47fb2f198145639f6d5cadd60cc6abfdba71d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.q_proj/_s.pt
index bf72c45226e71a8ee050c18776d7fa574545efee..ca6a4a93d0620ce31abc701650db4e8b0791e65a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.v_proj/_s.pt
index c038d98318e3ae3d3c0e83f323ff8bfc2c3324ef..62e21a26e87078655163985ca6a84d906d717f09 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.22.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.fc1/_s.pt
index d418650d60db917ddc5d61f42403db931c444258..4a496a14c05d441a598fbd8efc5f2e897cf2cf75 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.fc2/_s.pt
index 1526a59936c8455cbca9f06aca079cf698bb7860..3a57b5cb3e0140895a7a251bb78e5d0545afe4d6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.k_proj/_s.pt
index 179346bd1c9f998a89bb3e143226a5ef4c05b2d2..d731e252b0f412e570e940417040a613c6540133 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.out_proj/_s.pt
index 689e2c875051c8a8c190d542855933fa01b3b92c..701ffb2b7fec08e2f16cf8ccf3594a7b12f8eef7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.q_proj/_s.pt
index f72f69f929c5b74a04dfe7ee2d1a1c40d694d6ea..45bca106e123b42284c2daac1bd240e77cab4d51 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.v_proj/_s.pt
index e083068538b3f64a3dc745598a8ee35c2b471115..be3b480be4dba5a32a2997526a018ffae6383755 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.23.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.fc1/_s.pt
index 2f3566d1e4ae484f6ab30b99c9d1ba35bf39ef12..20a0373a8c42d883b93891a59c228eda3917dea5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.fc2/_s.pt
index c88b1c63b69fb7f48f94e32194db51bc3bf45a82..69bdb858607d0d319fb8f13868e5af77d0c30ea3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.k_proj/_s.pt
index 1dc3370b76eca3a5a665da2007c012e633bfdc8d..6ccaad0d923343008d30d5b63a7a0b68a9ead1a4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.out_proj/_s.pt
index 3ac6b519512a32922a4ce5482846e680585bc5ba..da28aa45736ce5f79b706a3c84e095c326efa204 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.q_proj/_s.pt
index 6e047ff037ddde5c5ae7f780f50d5fb29b513cbc..95fc5fa2a14339c441f8cd2ae1e0932c5dbf40cc 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.v_proj/_s.pt
index 6142ad242232141ada24741cbaaca63d182ac892..168524681db11624cb3bab74cbff8c8938379d36 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.3.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.fc1/_s.pt
index 31c8e417b7c52cd5a5954324398c544f3b2338d4..9659fb5b2681b4c74e810f3bee5b838c4617819a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.fc2/_s.pt
index 4ff328728f2f549f01c99a7e5f45bd7c50da1574..59828927187d98b2731acae13a02e060a93fedb4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.k_proj/_s.pt
index e5d627b3b2bb3769b865bb46531a30c8911f9784..5aad19ab7d09017b2760e732ed470cf6be2ae588 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.out_proj/_s.pt
index 92b8e0fa0f9ac36111502cef425a4e2fff105dbd..272814b174c42672c4675c2aad19f32d88cbd861 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.q_proj/_s.pt
index f0b42a741cb68346a6422249bb314b4969dedc84..dad9250bcb841ef7b00bc0514aff6e4034a4b902 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.v_proj/_s.pt
index 078ec78964be8c425aab5db83a6f77a5d4d9fd99..6ba70fc3730c1b5c774901cda644ad2917ef1d45 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.4.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.fc1/_s.pt
index 598f925f37f833249cb96d08df3b42ff9a0389c2..e150ac5b42a50aeac07b0dc2058044eb46eb8be3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.fc2/_s.pt
index 4ed290ea80a90dd4f1a5f34f456be6d6fd19461e..bfe8eef4395d099c20f46466c5531dc23e5de810 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.k_proj/_s.pt
index c5a35e9347e9304d2a731bda9b655e1f983c228c..95e6e1d44581856f1875a2e2aa2781b02de69c8d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.out_proj/_s.pt
index ed0eb272a1471d8a4518c42d45dce23150be88d2..2f61a42eaf4989a478d6c69787a80e88fa99227a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.q_proj/_s.pt
index 567fd882d24cffb7bb8063db3109aed70344d3a7..d9b8410476094763361c0b96bcc4942539d3b447 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.v_proj/_s.pt
index 472017e9547e9b8740ca639bd9627c8bd5ad0890..559a0ad7b2d9c5de0694ba7bf2f22fc11f57467f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.5.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.fc1/_s.pt
index 89c59f447a97225a1e532c775eba21c0728a9cc8..a8976bc47bcfa8390896204465d2d05d18589b42 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.fc2/_s.pt
index 9d3bc99b27f27da6095f6109c8dd80e2a3fb803d..87ac90e2b6d28ef1fb0f07961de8a4712941a2ac 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.k_proj/_s.pt
index 4c4e56e6a490bc8bc1c0d6eccaa43878baf482ff..816b014d81d6bd96ec2dab525b35505580eafdeb 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.out_proj/_s.pt
index 9bd64015b40b3be3a8b2cad23fc3a973c668fcab..db9b6ff7352c054ce3ca38d42318f7f098803643 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.q_proj/_s.pt
index 1d6264568276aabb1516a62bab73ba2c0fc230f2..ff9464ed166d7f344f6dcdb84a65eedfa3e84d38 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.v_proj/_s.pt
index ee96fe9532fccdd8a9862f1ae50799e9653b9b30..e7687c9ec88696d95b5f0346102e5052a53015b8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.6.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.fc1/_s.pt
index 0bcd82f65de751581f94d0e05574052ff68b656c..fca5bc878424f7d48bbf5b25d2a952d9d92f2966 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.fc2/_s.pt
index cbda55b0976e64ffe70b66acb3bf6925bf04d039..1fb385720ef243b3c31dcd514643b9834c8c84b0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.k_proj/_s.pt
index 999bc61b5b1934379e1f850c30bbd5c1f7d21cb1..2fce319461517958e5e8f787608d32099ac107d7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.out_proj/_s.pt
index 42c3abe2fd1ba44508f6c650f3d3796b0ce730f5..fc974179345b37d474a517aa121e66db512fe0f9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.q_proj/_s.pt
index ec81cd8fa6357b31e7d762f56af6245c5366fc7a..d470ecfc25dcbf06473fd7faaa84e007842ea860 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.v_proj/_s.pt
index 80dc5078a7291626c5d0ad54a8df3fd075daee89..ce13398b72af2443fe547abf293705922643ff78 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.7.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.fc1/_s.pt
index 60081e59f4616b5fc8f0f3f43e7df3b88036c1c2..ae1c806afdf6f3e7ed1e868e555ea42ae7aca411 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.fc2/_s.pt
index 25629ce1be84fa2f6b869d53ba7bd87a5be2a4c8..87dc9223a431a3d0c67aa9995fb809ade60dcc33 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.k_proj/_s.pt
index 8f20bb98885091c8509082ddf93d23f3546f70fe..ea0d20d039c4aa9196dca230d4915aeaed5ca733 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.out_proj/_s.pt
index 6859a1830894bcf2adcfcf26167be248d9f30950..7ab23e3bf674aaac38c9b76ecd74e3afc223701f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.q_proj/_s.pt
index d2d8f5c96a0da032bb092206412466a14fc8446c..2da912e315f466fef67d4b915dd08eaaf829c16c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.v_proj/_s.pt
index 1b14f26b92c8cce4c61594ef800adac90b920b00..3ce77489c83c41df023818a2a50fe9e5d79fe07a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.8.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.fc1/_0.pt
index 446eea5860a21090a5f7c7578829aa7ea1576bc0..58c96248fd00550cc15d645205571ee67c337035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.fc1/_s.pt
index 551bc92b61a6fb9778d3756c83071075c2f0c6f4..624e08cec285b74521890f73e8d97632477b1949 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.fc2/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.fc2/_s.pt
index 589add43d2520a4fcf7b9390aa41c1d58ce2cdec..a346fbd44518927f79cc6b4770df466db9cef19a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.k_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.k_proj/_s.pt
index c8f69c20db87950c8af4c0399292a69fdc6f7188..7c3d913ab6b3bf9e0ade686c4f538a94753659a9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.out_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.out_proj/_s.pt
index 5a4cf40b10a6cf2aebd66a75b922d2e31b5023da..7445469686d527a153f40593ef931392bd1ce699 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.q_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.q_proj/_s.pt
index f359c86fa2c12c1f558852d6e4a0ebe5ee177bae..0afbd26afe5edbc0772951cc132b8bf890d2af86 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.v_proj/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.v_proj/_s.pt
index 4be541afba4d0396edd09fff6df39bc5dbf9bddb..60deebc1c937d154177ab6b962e79acbc6dcfdd8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.layers.9.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.project_in/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.project_in/_0.pt
index 7a571786dfe022608d7168a415a64654991d2549..8ea38b2f931134a804d1dc0f42d8395e117dd20d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.project_in/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.project_in/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.project_in/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.project_in/_s.pt
index fdd2418f675ef09160d7f43cb81a2302753239ab..0082f76d68d54ceda4672c738a87d534a6dae46c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.project_in/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.project_in/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.project_out/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.project_out/_0.pt
index 6f3ba354ec876f170cb0943973ce2085a27e0750..7dafd7dbb653cbd2daae7b5c32e1db4207127f70 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.project_out/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.project_out/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.project_out/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.project_out/_s.pt
index 83485d97d9e19c515dbdc74b0202ba6b41009971..437f39ad3bad52e9d4f8871aac3e35301da5725b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.project_out/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/init/model.decoder.project_out/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/lm_head/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/lm_head/_0.pt
index b809e9c6a81326bceeb886e87867ed6f6d5566bf..9cc31638b5b3aaefb081a4f09a585059e913ba32 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/lm_head/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/lm_head/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/lm_head/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/lm_head/_s.pt
index bb18ea0e8b6e845d90dd4495e2545fb82019eae3..03f4cf02d17cd10557d9a492c61ba53f3a852dea 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/lm_head/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/lm_head/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.fc1/_0.pt
index f1ab16275706b6045efbd65efd87a58f86ec9be1..8f6ab98cbeb5819bb6f6533887320ffead5daebc 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.fc1/_s.pt
index 1c1303bfaba30a2a2f416a4ed1247feab6262855..47ef620e6b5c6581345383dc2641547f33ec7059 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.fc2/_0.pt
index 0b28f241ef00aa5fd20455e94ba175694ccaf3e9..0aafaf76bdc4b89d31619169b71d2202108c9549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.fc2/_s.pt
index d64c1b0c280b3d5d9923083b95b07a85265131f9..f169fd2d0de6e665037168f1d366c5030fafe5ac 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.k_proj/_0.pt
index d7efcad0af97dba8752fe0d006f5cfc8a96286a2..483d9cd7f21c9f77db875150fdb96555747fae2a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.k_proj/_s.pt
index 490cfd837391fae36506008d1a73c310ee29753b..298259820ec0674490722dfb2fa6851adfdf5143 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.out_proj/_0.pt
index 5b33f08f2fcce88705fc3b2a21a201559d08e760..4f07cce3d5364db1b3574312879c6397a62c3866 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.out_proj/_s.pt
index 1cff0ca0b3fa6a0d04a340981122eba01db8d8e2..c9a81ef4d8dd49afbc4837a0292d45915d4113db 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.q_proj/_0.pt
index 08adaa779fe7c2e8a392f3bd5928dabf8ded3120..b6c02c1b4927390c6d1622ca2e3a6bffb2e75bec 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.q_proj/_s.pt
index 3b574f2e6854b6d2bf50b5a8dcddd142c12f478e..e97460d2d488d82d978f2326c8ddcd0d113be846 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.v_proj/_0.pt
index 4bd8b275d8ad2d3fef6c20929d3c6445f7a085a1..38485c48a9232bcb6eda29e6cf470f247cf4434b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.v_proj/_s.pt
index ad57e17585b2ad68bb7532dab2f65c60db7877f0..81df0e4a1bf602973c327a62496218208df75786 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.0.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.fc1/_0.pt
index 5d11fa00f7667689fce3d04e028f6750b943d0e1..2053b560aee329b3c0c2d13c3c219eb689a9f7b4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.fc1/_s.pt
index 4e52b1adc98590ff69214296d9f5393fecabad54..517db1daacd121e325412b42ec6e487869939efd 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.fc2/_0.pt
index a0583500ed45f0dc3fc858adfe4f227999deeaed..e372874da37701b05640f3fd992bfb6a6ef83e0c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.fc2/_s.pt
index eca2d8e8a1692ea44d9689bdecc28ffb1d0549cd..cf41eb353f124b36983f0d0e166b30c8ecb1f3b9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.k_proj/_0.pt
index a8d870159eefc2fe4f44ef4125ac39e24406e4f0..45a295509a0c472c296b2d54f0c32c79348d993a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.k_proj/_s.pt
index 4588a2f0350e1d5f3e1a5c8fedad2801ba3f5144..13d822f1bf0f35b6da4d4911e2c3c486b184c004 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.out_proj/_0.pt
index 971aaa4bdec99aa06d6b5fb0fa3ab29fda102005..965e350cf257a3ec8b1a4e20152489636e62086a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.out_proj/_s.pt
index 9f423fcd5d4ecc0827699d6e879cf6cb6638fb16..de1303aff5fc802dbf5ed3b14411c85eadb4082f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.q_proj/_0.pt
index 71ff75a6c15488fcb96e96d0d12901145c47e03c..5cd3aa36f98a54b7a3b99d8adc5f45d6453d6da2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.q_proj/_s.pt
index ddc3f74f197c379035a11b945cdc5435caeceb7b..d84977f10868eaa5136c8e51e056e666481b6fc7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.v_proj/_0.pt
index 8a4c0b5349b28f1335a340e7d53756f37abdc9ff..e06fc752892136131cfeaf3b0c2c1b4ddd8452d1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.v_proj/_s.pt
index ec549f8d668efb3fe40debef253066099864cbd2..b19604290f444e66ef7c41e4d6d1ef45d25cdd5d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.1.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.fc1/_0.pt
index 97e5a8f667c1000d42aeca8cbef4f0d217b76029..0c468b48c068400e565a782d13bc1cf76ca89518 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.fc1/_s.pt
index 97303a988262cd8df53cd1c003daa5674ae30f76..92ef9c5306206dd92c24bfa99dd8011e6dddc03e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.fc2/_0.pt
index 3c48e829fd67a39991039f4212ca300ca604c30e..91e542fc5402d64c584ccc7eedebc2404d68b01a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.fc2/_s.pt
index fca769963c7a95a8dd7fe8537e7cc3e6fa62fa9f..b71a652600e72094839792724310676dacf68731 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.k_proj/_0.pt
index 9db4f1997d653c1bb1c1519627be2539ae17e15c..5b2cc6b25c5b513cf4bc95512c15806bc042e57d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.k_proj/_s.pt
index 9d2680c539d035bf9219548c8d65adb4c712305b..9042d09707396b6fbc2b6497cf7b5b35c2f913c8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.out_proj/_0.pt
index 7f01054bdc0e84837b9d141ccecaf7ffaa3feb5c..a4f4e86e92bf72a195d6ce7a8d70888278014889 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.out_proj/_s.pt
index 725f02a6fab7585b31aa290f2b8dc0b006d1b26b..edd291dc078fd60a0e4ad43f66228475bf4f2441 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.q_proj/_0.pt
index 1a687d3e6f1cf4d886571d190abff4c99084b4fe..45546ed89d32cad214bd8b7346ba083dc66ef62a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.q_proj/_s.pt
index 3634c27a81cd584201068bfdcf7e12090f463608..fd15ccc1aba48d76eacacebb2c8b951749d43ffd 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.v_proj/_0.pt
index 9902cb028b9e993d56c033fe30593984e58d696a..c81b84fc62a252d7b41083fc44fa17468a958fac 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.v_proj/_s.pt
index 4813a6d1904ca4d26a25caaf3c6e8a25c98d4573..c0988cf2cb74d9fb3e34af46d43824c9b44dc96a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.10.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.fc1/_0.pt
index 19dca6400b3a9f7f5bd691ba13d3c2545e2e12b8..9cda4a557e571a7d13aa30a7bcf7ba72543d9af1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.fc1/_s.pt
index d5d61fc0637921d94d85eeac278c08a9328f9758..33520ff11f1185d3a2262f9c5ad1ab2a1ec89d6a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.fc2/_0.pt
index 5140390ee2145d9b0000d7319a0fc8f92173066f..fefa2a99f0fb8fe1230019ef0cfc39b6bf556783 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.fc2/_s.pt
index c3787d4dca4701fff327c2c218b73ffb0bab5158..f1ca92162e2c5654cb549eb1d5a6a8c8364d75bc 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.k_proj/_0.pt
index 7e0b0eb4439ef8e0e8950a97158cd98209be99a1..08ebca52bd217e8633e8fc6ccc550bfe5ef2fb36 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.k_proj/_s.pt
index 82ffd04786246157516add5ef57525368ed0529f..7662da783656fb42027eef132a0a43c7b7bc5ce4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.out_proj/_0.pt
index 9c0ee0f6143811f4488b1522d58b991a2ea4736a..4474472708f25fad65907c1df3c147b4376fbdc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.out_proj/_s.pt
index c2796bfd17501b54f68f3ba6e4444ed12351779c..c3c4908fa6b62c65e3c98539029fa34a6d533573 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.q_proj/_0.pt
index 177ab56bc5e088dd9a01e4c10573501d91ce2852..9d8418d41b7fde875ee9db608dd7da593fe56813 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.q_proj/_s.pt
index a641b2c69078d8978f88a2961a7ee0fcc8fea98e..51793cc78b0b037e210238fc3930cd17789077e7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.v_proj/_0.pt
index 6dd0480297a86ba7129eb6a355ee01e5000d2e92..840bf7e6ecf73ee6b79c48058a339ce5865436d2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.v_proj/_s.pt
index 265423f2b8db1d536834dfb03ea93aaadbf17485..dd0011eb795752c9019d24a06edf042fd4f2df4b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.11.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.fc1/_0.pt
index a48d589d36737c7705533da470141f0f83efd98c..b824da233d092dd91e5d49a26c547042cd5f92a0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.fc1/_s.pt
index e21bb4ef52b8d2d119298ca85fd38f7c5c9209bb..90156459996dd588e38a102688368317d523590f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.fc2/_0.pt
index 969188de55a668af7fce2b341657e9702e1008bb..95f36767bf9f30b20816c1fa2cb643fab8d61321 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.fc2/_s.pt
index 12cadb51b31704573588d770eee450654afc509d..fd9c96939ef8900f919e1d0624f478bd790a7daa 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.k_proj/_0.pt
index cea7adccd28645fce324cc0965de26e19c7ec50e..3ff615348b7c7bf3be4713d587557c6430f271c4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.k_proj/_s.pt
index 699bab31e8681101a2a41ce6f8bf42db0a15e22b..91e6b9ad97b538da2a1c1d0b2312dc66aaf95f10 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.out_proj/_0.pt
index 8ef5220b56809d4940dab61cbf9693a355777314..ad6bd38078250b23647bd579a574c6a70bf780d1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.out_proj/_s.pt
index 4a4f1543d99e5b6b075cb2e153408b44219eac37..67cd061859aa6980da9fb3b63dc8cf87040059be 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.q_proj/_0.pt
index c1245ec79fa77cafc443240c7514799e40cd94bb..407f77cf1fb57470fbba2fbc97a5d5f88d85d4ca 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.q_proj/_s.pt
index ade345fedc70f6a1ac54f17bbc457f5a9b7e2567..a37b69a2e6e00eb9280118ecc1ea49b7432e9961 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.v_proj/_0.pt
index 9d9aee4a888cd50f51dd7368faeffe1c262a8098..01ef9a18d1faaaf0e030ca6543ab5c471f6f2065 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.v_proj/_s.pt
index 2b495eeeb0f2f7238ca9f45290bf6dee5446526b..9dcb2c6196cab407416199a09be5a8e9a9c21ab1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.12.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.fc1/_0.pt
index 80662b4caf0257162f7b58101b12ab9f1d4205dc..4d71fb4b898d774211bbc16f9cf991aedb209e41 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.fc1/_s.pt
index ecf449841560a5c17744d03b36ec7bd4de16e7a4..3753ea6707ab3c9b2f70d04250107623b1ade0b5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.fc2/_0.pt
index 5548c7798eb5e5ef09ac7290b36d9ca7d16eb68e..5a9723a81e5922e35355789250d9dd0688f030df 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.fc2/_s.pt
index 9f4c0c6ae13f2c13738d5985c2767b9573a25f5b..7c9219b93822392ed3afbda023e065db9c5534dc 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.k_proj/_0.pt
index 8f8e8048c74a514b3c1b5fc251a3673b1cac2cf5..1907ec359b66ab92c1336f650bf6782a727c6f94 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.k_proj/_s.pt
index 4292ebe9eec98a3658df99a0f364044cb51db4b0..92c0e32064e52bcb5a1cae9d3f499367b4054063 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.out_proj/_0.pt
index 55a35978d5f525b24a87575868a1f15b725ed2f7..7cb790de5df8e27ad76fa3b3cbfc6cc5a5a28cf3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.out_proj/_s.pt
index fb17b65f35c6381928cbeddfb06bd571c17853ae..4fda7866a4cb5e72c8dae7d9a9724b9c52a8ec75 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.q_proj/_0.pt
index 14978306eb02cb01e5ae8e3373d47e74b15e5b35..e983e559769d3c28aabcd140067191f0c7ae2fdf 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.q_proj/_s.pt
index 3993015c11f42e4cfe112c4dc476a2b21c733542..dc32a42c72ad4be9c5b9d1468fff3c2311802df5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.v_proj/_0.pt
index fed95bd0f966bf9cbc08b12d13be757fe7c1bfff..b580d89cf9b6622db1ec6902359f883b86d52986 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.v_proj/_s.pt
index 17909260c25cf12eb3682bf333beb928b1f31076..67ef6fb2bc6ad3dc58ba6a2e6a1a31bfa30ab2af 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.13.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.fc1/_0.pt
index b9007704bfbfff8fe7b06af5bb4cfd9ec5760c42..452a3cb1c7b66e0e0cf2302d4244e00847e25259 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.fc1/_s.pt
index 5dd5ed2099ace357624faeee093f9236a3907164..fe792815211afe637116a59249a5c95395433837 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.fc2/_0.pt
index c0c1d2249bd2c8ae3911c0f748ec15267276b55a..b3d5b6da78065a2ef68d772f4dc9ea589349e284 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.fc2/_s.pt
index f76d2f2f8750629242caa8739d22a09caa2a8476..bf075fc52888c6a490dfa41ba96c1513f852e4bf 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.k_proj/_0.pt
index 3b82e7dea2c93016325c48851ac335ffeee47c07..bd4d4211370fb754a0114586929e22acce93a672 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.k_proj/_s.pt
index 6ec0b2dd306e2f4bbf524d8318a3c983588f8aab..e65c5fffe1b463b11f10e2db844e0db55c6f1cef 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.out_proj/_0.pt
index 3260047fcc8202d9754f4ff0e186b44f26820182..9bf7a447f35c2362e8d18bca4773fe240922d1b0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.out_proj/_s.pt
index 9e481d820f5303dc1d13ce6a1b195a213dd382e8..a53ec56e89a99df9af61ef1b537dc598cddd1bc9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.q_proj/_0.pt
index b8b12c7a86c61423911d15fb911b246552fb73f4..61168050ee94adddc3c2d54f77830646bf3e0cef 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.q_proj/_s.pt
index 3b7276f8c53b64236f0d18d0890f3f2b0b5a8452..3cdb8afb465607b9be9a77d4d8924341f0d341e1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.v_proj/_0.pt
index 27057be4b5b434e0b1f6037125b61361e45e82e4..f2e63ec9cc8b050eae5084c7eb2447eafa3bac66 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.v_proj/_s.pt
index 72caac6f1ad02562d4aaf2cb79edc3f2b808e308..0268139d5e1e97ddfbb1a9a188c47ef33b236ec8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.14.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.fc1/_0.pt
index 553011f6fcfe82a756a1bd554b83cf3d05cc43f8..1ac6d7e48e77028e2771ce924a2c839095946960 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.fc1/_s.pt
index ae634ae1d7c305c345ecab563544689dd22a0b34..182ab25f73f2165c537b16a5ed37789d36d2fca3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.fc2/_0.pt
index a8c308fc78d8d17f4d9e8f9dacb51f8eaeccbe9f..47dac8507a221db9594371cdcc7badde135ab8c5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.fc2/_s.pt
index b2ef2e1e7498ae79692aef038d434287bc1f33ee..a172d9989dcaf16d7054d77001a8310a15008ec2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.k_proj/_0.pt
index 603049c0edda735b58b0f44bd3c46cc38e8ec517..f30af30ddc40b25a2db510dee7fba722c0e39d66 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.k_proj/_s.pt
index 4ed541de2a51186e9481ebbb69bf016130eeaaa1..ab0d958932809e6dfa8d6c1038ceec7615e6586b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.out_proj/_0.pt
index 1b4905dcfbc48f25edd522bdb09ee3afb4a974d7..d231c27e846c909a9e323ebc53d8801aecb228a9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.out_proj/_s.pt
index b69f2a30e23b686faac27eface61af4d77e5b848..bc12d5c3bacc580aa523571bc3f44727c610348e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.q_proj/_0.pt
index 4dbec141359357806b0041ad006ef1bc2fcc3f6f..5b245491dd283abbd927b71e3e2d9d47b7e9727d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.q_proj/_s.pt
index 8e7d61a602b8b3592a346393a5805788babdaedb..a00d4558e96c3ed5452c6c533f4f8998dde0629d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.v_proj/_0.pt
index c346feab1c0469c08b43856faed54880eef0c25b..a651f7d7875e9bc1bd266e21b6fe47b3393116dc 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.v_proj/_s.pt
index bb03a627ed1f0af74f6d21dddb3a96b7a0698986..22ce241943030c9ad407d5c1b48c0f179b2d68b1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.15.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.fc1/_0.pt
index 4fcbc9de82b47dbf551c1a302801edf0da2ddd60..259ace69648416aea94e73120a73c70fb11e4bb6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.fc1/_s.pt
index e027cc1f27735f5e9c7199a09fe0fe8af61b284a..c5dd82ded90b82e7a0baa4dcbc13605467d4d0e9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.fc2/_0.pt
index a7d1651b4474568b5c44835aab7e84dfd011c282..7d3ab15d3e54d7353bc52f9a610d94adf51ec0c2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.fc2/_s.pt
index c71fd559b6419be5e3ddb2c3eb56db441638dd25..9f9c05bdcd2eadbfe5f5f5af05c038d964a700d0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.k_proj/_0.pt
index 80464c94e548dff8c53c4b8fd01e92583185f0ab..2c425a2eae8f5b8c3f8f0366a264dd0191408f0c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.k_proj/_s.pt
index 54e7981a7f2d6602aae993f0ffcd3a75762403b0..224fcdb8733726bbd0c001303f4dd6c9575ce0d5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.out_proj/_0.pt
index 400ce3512c5f792b80141a89fd7693d2dee554c6..5ae2af9544ebb0b4637a5bf23412167366e92b1e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.out_proj/_s.pt
index 4ade58cec672833f32a4bcefd1cb422857e9734e..e10e739e585604f410b06a812c2cb328fb06cd0b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.q_proj/_0.pt
index 9149e4841147e2073132231ca9a4e2b58ee78c31..c711b9db8258771343b42717f21de98e5ea67745 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.q_proj/_s.pt
index b78206bedd6266a8fcacba2278ac0b5869cc900c..a29c56dd471b0cd908824ed84e734ae0e4ab487a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.v_proj/_0.pt
index 3cb234ea48a1b8f245fd181342eff84878dd105c..79006d7ee5c4274f07b27d3f16864d42192a6f09 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.v_proj/_s.pt
index 8de124b65969ebf5dcfc68f24751031a0625bb8b..03b1c9cf53649e128b274cba41f0043dc4da48c6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.16.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.fc1/_0.pt
index 68551d60cc114d68bbf984d10f859ddfa20132f4..0f58bd9f7c52fb82ae7e4bcee3da53ca00c39517 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.fc1/_s.pt
index b4b07bd88bb4544fbc9aba58dce5bc400b4d6054..85993c8bd4f310d9f990f5004d43a4d0f08ff3ef 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.fc2/_0.pt
index 5bedc8e7c26ad51d82e966694fe50b469092c093..69a1df53ed20c95a9c857b149715635efbfbac90 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.fc2/_s.pt
index 32a80464f4eed1f2254e8893d418fa3aa8dcca9a..8daac83bcef8044187fcc1606eeda2e150582d08 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.k_proj/_0.pt
index fc2b81656bc62d1f048a8289742076dcd36d040a..e7b91a78a17c07a909ae4aa368b2c670341b4fd4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.k_proj/_s.pt
index 621c3aca6022489d33ae81fb811da8fd6707b15a..ae0c278019ea44b0fb02699c376bac6fbe31ab2a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.out_proj/_0.pt
index bc508f5de80386f21a923fb33a7258925c325cfa..1ea901bfe8241b6c6762351bf5d82bc6be4f4e27 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.out_proj/_s.pt
index 7faf5ad555ce2a75c83a635fcfbfa6a6defc534a..1f64a62523ec210735518c60037f6519fc519d99 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.q_proj/_0.pt
index 81cce4c60cd0c54e9f0d6936e1bbcdc9dce04ea3..635011ca6a2fcbac64218ac4de4076185d49cbc5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.q_proj/_s.pt
index 23338c3ee69d9bcf3b06101bca5e3497e5225dd5..c9a0ea4cde414a2f72e1638213591a75d172aa5c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.v_proj/_0.pt
index 3d528a7ab91e3849b4227364e8174c683a8ff463..cfcf36cf8cba4e91d77f63a406af4a5fbe0318d5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.v_proj/_s.pt
index b3c0e3d391e0a86a6fd7f452e10f0b0e8408ee9c..748153f554c79b62b9235d92894a77591f6be4c0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.17.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.fc1/_0.pt
index 550413311257a49c15d5930c81c550548fd6198f..409a221ba696764780330fd68dd7b735e50c7e30 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.fc1/_s.pt
index 12165cf65b10e608d129ceb93181b16b7fd335b7..4e422a63588d0c04e3827e9fd7975cfca65b06ec 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.fc2/_0.pt
index b4f7ea7d39ee8fcc39bd9f65f0c0f1f10a7a15c6..b935c13321ba07aedff6bdfa5911f6a5f000aed3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.fc2/_s.pt
index f409716bb58918497957f7dea71db3a7880e6c12..789c3d542ba85e267288bccfb1e727764f24064a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.k_proj/_0.pt
index 411b5f6928f6a932fd8e4eedc0cf3eef17702bb4..e55c17bc8fadb0be966dccc3aa51426a0df8fefa 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.k_proj/_s.pt
index 8bf3f1df24934c7a9a72375df6a28d0a57fc41b9..9ac46ab4856f110fd2057db18b6350f7f0523cc7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.out_proj/_0.pt
index f4d01dfbc2abeefdb5e8761839ae46bdf909aa37..7a5399b84828233133ce6f6360cf25fb2980e3d8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.out_proj/_s.pt
index 628db392af06a14ac5e78f426ab750bf7a981551..4956f0785732c72df7bf12b5adf4181a82b6b9b8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.q_proj/_0.pt
index 766671e8304386982c828698aad5a808baa0e80a..e09ad20633ab4b24c31e18ae06d2ad6e3e7b87e1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.q_proj/_s.pt
index 572262b170a6bbb8632c71fc030351946b88a442..393437fbdb68298537fae735adeb98c33b14bb9d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.v_proj/_0.pt
index 67980b32c488c3ff5b004b8dc64ff08d6a8bf91c..75b6e7d6972e1b9a30ac14931c94eaeb4b8efaa4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.v_proj/_s.pt
index 91c819092d8c64b4b3a2507826a04c2217f5fadf..fc994f2a2a2954eb7187bc2fdc31c9768dd72ffe 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.18.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.fc1/_0.pt
index 7e308e5a1deba76dbae5af4d144aa8105d325e07..3b2ebbb5da24eddd9c0fdc84ca0fa998a891f11f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.fc1/_s.pt
index a88a5d4bef55428e4b05f90fc2e97fba862df974..a3996ab334492689862a640209e75d646abcc019 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.fc2/_0.pt
index fdd9d1c22b4f4db9dfb991d31bfc534505591eb5..dfeb099dd262f5f8216119dc63235523f296bf50 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.fc2/_s.pt
index b152b29f67dba38c1d9d359f5986795efffce20f..199b88121cd24d4ab80860005937d90b13118fa9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.k_proj/_0.pt
index f54c939a49525ab27bb2c013b46cf3834115255a..108edfacaff80fa1a2eb35ddb0575743c6637b0e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.k_proj/_s.pt
index be63d0fef57c59f307ad6f2361956e32f24312ff..2a0262e52868456759335dd3f363765d46b44f00 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.out_proj/_0.pt
index 1ca1474067dffa71fd7d46487995ef36a450e80e..5475f6e62ce512e799b843ea35d7f8cbfac09c4f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.out_proj/_s.pt
index 6d6b7d2914b9651cce480f5872ad86fa107843e9..5f4a85bc8a573bd4af4fbf38afe5c24d205e9d19 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.q_proj/_0.pt
index 64c2ed98581f7840460012b2df758c0085755d14..eba55212e9120d15997925acf5dd07feaddb4507 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.q_proj/_s.pt
index c802f6ec2f82eba368befc92a02d54f8f8158aed..37686a83cec5bb52a6fe8f484045c57dbc84a7de 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.v_proj/_0.pt
index 7eea75172e2e2ae23795575fb1e8581cf397f180..92c38bc2996b9535cd186c9a99b39393a67c164b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.v_proj/_s.pt
index 3ecf66887f3772c66c9c77f6bf2378e54cfaac20..834cfd878fe6e6d2aef6d6f863a2b99967fb499e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.19.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.fc1/_0.pt
index 7b17f4cd6cf9e08a65f12096a49bf3336d314718..f687081f84312e78f55a1dafb494a293c48cd91e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.fc1/_s.pt
index a5194dac6071ca6b1277acc62a2a2e7e125c099d..631dd000198739bfe123226e46ebed8140d69978 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.fc2/_0.pt
index 93ce5a30247198f5d7832048965527c8d0268065..240ae1a3f6e64a03ba6e099017b2bad5a91a464e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.fc2/_s.pt
index ea987ba6dd1c1d561ad627c94fe96138cced3d52..df607b943c948ce5a87f614c483338195d205171 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.k_proj/_0.pt
index fe7394e5bf75f5af1f29689cb2c5adc8b5e370fb..76f25c1ce2b60784d443ca06d9f8565c1b87f232 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.k_proj/_s.pt
index fa9484c3ddb7b862d37cafaaf83edb85e247be21..3a252d29847ac5161a6a438f5a14581098b676e4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.out_proj/_0.pt
index 45362793ad3eaf85c5199d53c878aa2d47bfa8b2..db96e14b4bee4cd674009e961304e2b9be4f62f5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.out_proj/_s.pt
index 53439c39daad2b82a673d66f9a3810229996e87d..b27866af39613b0230de12ed732e1caf9a28d2c3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.q_proj/_0.pt
index dcd6d4ce666a68a86b61f371b125e4fa421603c9..e2c31fefdf5462a8177b5dffc92fe7ac61032145 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.q_proj/_s.pt
index 366fc2ff8c48318763a0a33e23796db9546ecbf1..bc9963da8745924a1efc91eca6c53fd42c7cfc84 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.v_proj/_0.pt
index 52cba77482f53824f50b3abbc7a9a02dd46cf28b..1df54cede2587bbea12ecb3cea54247b5a7c5fc6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.v_proj/_s.pt
index 0161622b1994f9993b3e29899b9298a6ec6e45bd..462ef66a60993b80f0a81389e76b5b367b99ed9f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.2.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.fc1/_0.pt
index 6dccff817aa82ba404b03882f0875e7629b8895a..6fc50145cbb02fe56cf478ea1a17e5ed4a1276a8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.fc1/_s.pt
index acc16c93d8b6e5ce90c61759b215a78d3de62487..6dcffb6f61790181a93962288e081650f8388ad7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.fc2/_0.pt
index 8fde10dba2bdb787ee8779d8520540035354aa85..3dd63912ee1c99ca0342dc93856d7f7d85e2e11d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.fc2/_s.pt
index b00bbae2ee3d559454100b229b99f81ef224b278..a4587d435caf59c4e165abdd57b16a809c93b74a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.k_proj/_0.pt
index a2c5a02d7ad9ac39463f8b6c637ade689be9db29..e8de2b2345bf4414fd591d3b995ff1ddf86f512f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.k_proj/_s.pt
index 2b3c75ecc70de6bfd5271d5421211148b52d6b3a..6e8788d22d118d7aef18493f65ed80cf3b283394 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.out_proj/_0.pt
index b5393db8f84d6e6ee406e745d8aa5781ce69ad94..a51313cd023dedf9e2f840cc7200fd2c25e986be 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.out_proj/_s.pt
index 688f615bb8125d6777b24bc20bcc9bbd909270bc..1bd565fb96a7b97e68c831549afd70eaf54def53 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.q_proj/_0.pt
index 7dc4ba18866ff92ac58d25a52777c8e0d949a4fe..8703424a00c628a6d5598d3970b350bd3cf274bb 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.q_proj/_s.pt
index da97ad15e44fe00a4b2fc7d89e6bf1b1fc3eff77..93eb2719aa4b5b9b0ab92fc3fe3352a71c451ea7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.v_proj/_0.pt
index 6a8fb174098dae6caff5086b6bc9e2ec2074e349..1954d046191c4fb381e604d81753437cbe4c7f5c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.v_proj/_s.pt
index 01c37555efd161ff0a35141b959facfeb1dc7d9a..0d3ef453a2d58c58c4604c013a6d95bb9a585d27 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.20.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.fc1/_0.pt
index 60bafaf5d3ba65572dbfeff89d980cddc4cb1a41..d534e575fd79f9af21bf212ce138f81396b170f6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.fc1/_s.pt
index dc2f030f8855281ff3befbfa7eff3fb72ee152b8..620f76d8791039c5de9ce0fcf6661e7ae5d1172a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.fc2/_0.pt
index eda765c9aa3c8aaeacf0abcc00a6735a6de346d7..9afa72c82ea9ccd92e7cbf4863b490b1e98d575d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.fc2/_s.pt
index 4c8fab051d4fe7840c7ee75902ddddacd5839656..a5874c7db407ada03f269770876514c5edf41104 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.k_proj/_0.pt
index 0d7c96c371e93428e956557b5498e5a23210db61..70b839b05e688243209e844ea7f67f525e0955c0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.k_proj/_s.pt
index 8735a8ba7d640042a66195ce426b86f8d41562b7..46ce09a7e2c4f7381ffdfe191edbe12d9f5b8f12 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.out_proj/_0.pt
index 1c5baf9ba0cb6897951b9b72e3991c8274a4c343..104a87634293009d7c95fdb0ba62526d381be6de 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.out_proj/_s.pt
index 7b23842271ba3b6d93e796cddcd413d4ca8d49a5..4afa75b206e6f2306a26dfdf473e63ce750df02a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.q_proj/_0.pt
index beaa3fa70d1eb83fb8a1759c7e7ad8821d9fdda7..2a673afe03dd7e6f26f18392f0e18103796199fa 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.q_proj/_s.pt
index 662460722acf50556fa390e1c85dc06765a4510d..9d26a2776a044718c38147edcc42fd407f2a5052 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.v_proj/_0.pt
index e9061c58167362f07b8a010eb7d0e87e51ba4db4..5ad4357313989ef5f1c2540e0aea6c423dd82f62 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.v_proj/_s.pt
index 54b251660421707823c33b820901a7a3aa00e7f7..d9c3a6fab47442470ec3f4234179e4e34fc2ebfe 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.21.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.fc1/_0.pt
index 4022844485cd52408fcedc9ffdba1bfbe0998159..d0b5b2ae42d3778e11b1def51747e9aace6f592f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.fc1/_s.pt
index cb66eee6e2f5072b8c5da2d97fa63ceaa0ea41e3..b2228c5c49303a77a32c953d66c33dc618f6bf56 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.fc2/_0.pt
index d4690647071d6e52e73ecee19ce3c102fb5ea7c9..98320d4441ae995ed9a05017953406a72fcb15e5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.fc2/_s.pt
index ef2e543ad0785e953e5d7bdfb5b898fe9026de63..dee30a44bcd3823805de5543fe8d194677fb21bc 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.k_proj/_0.pt
index ef812e2a8a91cb4b898bc485a8d7243557e7ed72..c0a94a9d2fb5c212de2061edfef067a498b7cf9e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.k_proj/_s.pt
index 0dfd176a671316daab7adde1284004cd5c1a2843..242863b6911c62b3fee6b3ca0fbe15ad439e922f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.out_proj/_0.pt
index af539a4ca97454df74608ac28e7e77ce508e1d7a..7b1b78f7c39491a844318df4ea946c938a74278f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.out_proj/_s.pt
index 3cc27d352c4a0ac24ae62e86417eac50a0337eb4..9cdcf5abf29b3110e8a482254969eb825373b30f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.q_proj/_0.pt
index c9e27be55945e064d09509fac52a9a1e6a3bba84..45eaf9568855f547c05f2487f4b5bc9f62fa8a00 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.q_proj/_s.pt
index c5762ebd09fed9b781b0505dcd4e5b8eed09b30d..40ab25a6cc140b9dae10626ec4e6378c9ad0feab 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.v_proj/_0.pt
index bf2c4f9f4464cf3a2afcfe3fefe92e7929830ca2..1edbe4be9d7f719f6cc8a6153b27e76c6857a3c8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.v_proj/_s.pt
index 99a2fccdaaa1bb53e23a4077e0d59cc39f946c3e..fc5328233c5487d998935303cf345a045284cc5a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.22.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.fc1/_0.pt
index 6db821a92f7f1e54bd7bb7fda9ef08e343728e42..ba5676ba9600a7c89d48bf5e0f48e0c9c9d96f9c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.fc1/_s.pt
index 9d2dc63e9f1ce723eaea9518a601598930874577..a62e3c263f5635cb5502e6875a7cbc6d3740c509 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.fc2/_0.pt
index 34d62148cb0db3b1f43d356f156136e83f379a7d..312827832bc56ddbcc3d468e83a3570934399494 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.fc2/_s.pt
index 2b8a09f8169a4d53e0fdac540f1ca99ee94a8dcf..b3d53b1fd03794f4bf4de53ae3fb5e2333c75251 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.k_proj/_0.pt
index 740e1493560c719ed0e1b9aeb5f18538b1d54e0c..6798a3d1a0850e4d49c335bee911cd4d849b4baa 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.k_proj/_s.pt
index fecfbb799602e7e2eb02f6321d74330bcff8512c..043250317df9488223a2535b585e5dd556155d95 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.out_proj/_0.pt
index a3e5860e52715f3ea79196d6d93ec3c05235d1bf..5b661c37efa8cd3e86d3d077d7861bc6be746f07 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.out_proj/_s.pt
index d3779b7aa9b47de3f510f7dd08f1fcb79a9bd83e..b49e1520d7303f2817e62286151222327e4704c9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.q_proj/_0.pt
index fed10fd69c105b1c461cc1bb5b56b02b645fd5fe..3d2ae1ad3f635b9c262baae27cd0dd14e69d9f1f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.q_proj/_s.pt
index 086f0cfb144a1e89fd44250e170a5a0f821ab8d6..5b695a66f68b4cd05ceb7cf241d0be7110164ad8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.v_proj/_0.pt
index e04f26a4183a3952f18963808b6661dd7d8a6ab8..fbf3df686fe62748f1246a8118a7de3d4f5e927e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.v_proj/_s.pt
index 82674a19236080890f9e5784b14609ee2c4f611c..9c897e3ea46debb2cbda36483f86c4914880233e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.23.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.fc1/_0.pt
index 46a303547bf99e2bb95da9577c81dcff28339665..74021a7a2226fca74159e4899d3379ad5c76984d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.fc1/_s.pt
index 22ce661d5b35ba1a3608153884529c795e74f653..a568bdf2a08a6ad6a8dc297172dc0a9e9ca13570 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.fc2/_0.pt
index c79eb86134675d5a5a4c25e55f58d83944a8338a..0df2859c588af721915e8e7d7316acaf1938d3ce 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.fc2/_s.pt
index ebb2b89034ea84314c24c691f45d5a79baeeb2d3..17118050e780c6c7b446bcc2510e085fe3504d6d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.k_proj/_0.pt
index c2dd2d4fecb16092358a22aaa0852da2a26de44d..b9adbcdd9450618d12b6e5ef30afd3f9142461c0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.k_proj/_s.pt
index d59a113568c802679099eaabf3f6fd8b127c3032..8270898c6c770964d60d9d6effe56130b5eb1e00 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.out_proj/_0.pt
index 8d13b699330d92e32e300a36d230afa146098269..f24f9bcd13c8fe59df4708599b98ad8aeb5914c8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.out_proj/_s.pt
index c2597ccf5968ddcce08d899eb6bc06ba77f712a9..6b5230241a8723e1c75bbbd2aebb40f9e063f333 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.q_proj/_0.pt
index 93f1004541e6d0f6aa1156e05d99a1b98c6b05b0..b8007071b78ed6b918d984c12cb491ba852f70ce 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.q_proj/_s.pt
index de1468913ef0284ddd421cd267879208f0f5b885..ea71f21b67fb587579127e53a5a6c75b09a23673 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.v_proj/_0.pt
index c2e0b658838c011caedf3f347c3eef02a792c277..5fb75e2431631d458d265a39c021da96657cd07c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.v_proj/_s.pt
index 3455e3f3a44d324ccabb2e542f8c1768b0599245..25c520aeee690a199901f2777b3f3b81b85e8e76 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.3.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.fc1/_0.pt
index bcf17f0370cfeaa8596681fcc9f2c4924a697a73..20864aa7f3d9fb7b82d788986381df3b770007b8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.fc1/_s.pt
index 3eae4f5d0d75ad2e9ccbf8c17bd41c31d1e647a5..b6513628cfcf8fd9fa008f77d4c8142907d8c573 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.fc2/_0.pt
index 5cd166bf0938bc195a28759c7cdf74c8482b6e0d..b869c17f8b709bc6c578ac8185cf17b9dce68006 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.fc2/_s.pt
index d03d52049648d176cee35ee4d4fe0b6b629a96d1..995e045ad0439bbe7551d27a9792d12d1e9aa411 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.k_proj/_0.pt
index 2da3d39f7c54fca68282d7a002d0bb4919e7ab6a..02ff61ec9ee841a4f1fe5bcfae67fb261f38087f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.k_proj/_s.pt
index 6864fbac58054e0d06ed2675b1d4ddacd9352413..15f9be02f0adcd60ffc84c0399134f29530faf60 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.out_proj/_0.pt
index 394b1a69a7335f15ccd86449c1611ce37cb7aba9..7b4d87b73aeb013df5777e9b95bf68bdd59bd752 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.out_proj/_s.pt
index 3a42b94f0c15fc0c94871deec1646b4990e1753d..1c37b2099e96923984ef3aa791e99c7ecf26b2f4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.q_proj/_0.pt
index 3865218dc29ae84b45beec15b93f43cef175afbd..c185e479ac1934a74550432415e75478e63b07de 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.q_proj/_s.pt
index 9e9e3c38b83771050722f7ff7843bd5ba99b67d0..0bef0a9213a2949677344bb9eb40e525d63cb61d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.v_proj/_0.pt
index 5af6b0a8511111b5443f593d8cdfe16419089037..bea453383fc54bd021d0a7e2901b2525f6729c36 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.v_proj/_s.pt
index 4560a5e2e60621e4835938445b4bef3b6c5340fe..8be20680d962d207cff5f3ab77f58dc90d5f7138 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.4.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.fc1/_0.pt
index 4d90a3e370f4113e04b38d86256f569b211548d4..e7501afc10545e23e689cc747b6de08411cd5d18 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.fc1/_s.pt
index 07682fb379ca7b218a3fdccaaeab24515bc68114..5cbf073bfb28b2932bcc96aeb747f5721c6bbaa3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.fc2/_0.pt
index 7ac60b186641e9306bff66de03af940ea7124d4d..b7619b07e6664d9b88d0374ee68d5d8b59adc423 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.fc2/_s.pt
index 1a6819fb08b5b9ca5bdc367c602567509fcde15d..f78843b60ce3e474538230e4fe0fdd1530f7bdc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.k_proj/_0.pt
index 6c5e00c0602a16e9b23430111a6cb8647ef4c16b..8c556c4dfd7db465e748f72dcd70779d3918ed89 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.k_proj/_s.pt
index 54e4a269f903748df633dea9a2621ca7df51937b..e550b856d0cc0baf735e542e367fcb165fc47503 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.out_proj/_0.pt
index 633d79ffdfee341729551d973064c713a52b1ec5..0426918013748c435f72737113bcf1f771a78115 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.out_proj/_s.pt
index b1becea3b37c6973398726a6f922536d077bbc46..152676dfa5f117d37a267e7be4682925ac9330d1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.q_proj/_0.pt
index d0a29656ffe2d9c3247eab05e3780936da0eba51..40275bf4c3f56d4f3ace84b73b88bf128afa1a8c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.q_proj/_s.pt
index 812992dea865e11466abf15db74414737d841be6..1a634e7bfa918245b3ca8a2344810359d092c245 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.v_proj/_0.pt
index a71e1a99f7d47be83013d2fd15daa92bffec5bca..c464945ed9a0543551c3565cdf878ca6412e3b64 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.v_proj/_s.pt
index 844cde19dcb9830f26ddbc018dea2dbe105b8c69..881a2ab5b241f387df1c390d3835991e99d87386 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.5.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.fc1/_0.pt
index a089bdea8fd9f04fe23d624964ef9531f8c4b76d..585f69bd0afbfeb91560c15c3076f125301cacf6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.fc1/_s.pt
index 0e2a9262b0ff92fe96e76874b1d0235fe666e57e..c34ec3578e4fa05c06bc2fee878654f9333fa8f9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.fc2/_0.pt
index 967b78bd2e3dbb5ab4fc9868eaa09f58d77b461f..ebec8c5ed3319440431d894e975a21b70179087d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.fc2/_s.pt
index b90c7777d2e8a1b357ca70792964915a98aebe80..e8ac70dce3b1a44b0910d4612f77e91878cc8f00 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.k_proj/_0.pt
index ff83764cfc03a21d5eba891f6a01ec8af0337189..8ce01470c96a63a7d02a2ed6962ebbdc5831eed5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.k_proj/_s.pt
index 675e9e5617f65b3f5aad6a82768b0493ca219e00..74c87c7f74f2baf82e0f770b975c091ba8b3e787 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.out_proj/_0.pt
index 1d5b40ae6916cf87bdd93bb657d175eb53799c28..7f91fd77b27c5f3031746cddfe13a16365ae1afc 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.out_proj/_s.pt
index 1e1c601e0d32e9a18fbc30de022bb70a74c4e8d0..4b49fa568065ba1f275741902319236c062b28da 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.q_proj/_0.pt
index 92c7e8e0a0d4033fb54c45a9eaef411249ccc2e1..1d0b484ad94c609cf8551cc7e129cfd660bca8d7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.q_proj/_s.pt
index 05d2ec0e05b0c6f6346cbfa843339ee478b44daa..43c103c9cac2502e40349935d22518363c4fddc9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.v_proj/_0.pt
index 789f35fce5e47647b9f548c18d5156111dcf6b5a..e48c8e055275c86fd2b53f1a2c1a49d3f2a3e4a7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.v_proj/_s.pt
index 963be265a7e5c8bc35abda7e01f2a503750625a5..697e421282a41b67863c21462457decfe02de985 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.6.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.fc1/_0.pt
index 92e034440e8f3fe0255ae06b800de672711c50a2..a8c38f9b872c8ec103345ab3d8e547370c328c09 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.fc1/_s.pt
index 34b41356a853a68b63ede9a4ac41acf1cead8402..77ac3899fc79040dd33083b07404b5257fe79daf 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.fc2/_0.pt
index 6ff152ed6102210ae24a69b33bc41c6930623270..ef0aadbd0c290e0d4e272a3b1687472b78d0d76f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.fc2/_s.pt
index 080969ca4b9f6bd60be83fc27b2921c2dd56c92e..2385be94b8b98f632662a2eed39dbee70ebc7b80 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.k_proj/_0.pt
index d5393e9deb8002f023028c5fdf93c0b4dbcaa468..7c1f90fe808ab124ab72e4f772c29fa9bbdee6f9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.k_proj/_s.pt
index 19271998413b5a074c062609c33f7770a9fe6f23..a76ef586821d2121492942f4ee43f00f27060d52 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.out_proj/_0.pt
index 04a32dd9e67d3ac0dd5512fc9aaacd148681f29a..b142d569ee7b61e74372b5fe6a12a2c9bd8cb05f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.out_proj/_s.pt
index 6820809bdea3e21ba7ab486d80bb72c958f2c0e9..025fef26620691b85eddeb1e6e4015f127c7a246 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.q_proj/_0.pt
index cd3167ee6b4539f74a7c5a66da7f9b1888f177ea..b86b63a142a7d43a9605f5ae56a7d79bbfb84410 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.q_proj/_s.pt
index 7b95a7580d0f67835d254781a1af8d67f0c80b91..fadf2b1744e4a4c634971035f205d4331a74f518 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.v_proj/_0.pt
index 76f2e88abd313b96981454815224c7953f5c9869..5e96fb0db6266ad932736f8a8ed33cfc89dd3af2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.v_proj/_s.pt
index d152dc1a252c247e132e341db4b384e24631f679..c8c3b9c9b49b41a0d2ccc3740630bd70adaf39b3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.7.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.fc1/_0.pt
index fc496c2eefaaf58dd87cb67bcda8bf827467e428..1501a76616b68b9351a1132ee06e446ade8c1113 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.fc1/_s.pt
index 18b6e2c66946dc8688ac9ab95ad683748f95c0ac..8796ce12e6d3887a43cbd3b085fbead65ef3e0b4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.fc2/_0.pt
index 41d249b69b7da2ab03a8f1cf633007f93f5fe03b..b8da111c0ccb26c11966fe3694585254ab77a69d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.fc2/_s.pt
index 278bf998a20b8ffff4d645942c6089d278adae27..06049e8e3513fc52a8dbd55bc51e1fa86ccc7f65 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.k_proj/_0.pt
index e928b6acdc9c893515928f1da3fd00d58e8db963..b706cc379733a5c3e6d5f324a8f351e388d705b0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.k_proj/_s.pt
index 509d0187abbe52d1536b8aa715ac5ae941f197e5..50b41894bd40fa7959f15ba0f1b2db71575186b7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.out_proj/_0.pt
index 29cbc42d963959f8868ad8e3d2fe47b6db1d0efa..d86b4baae3e7af63fb89674af0d4a65064499977 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.out_proj/_s.pt
index 11a258842b52f0cb2e2915ef10e7ad2e2f6214cb..81bc8cb51ba3b6cb95b2274aad7b0341f9659b58 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.q_proj/_0.pt
index 99a3198c9ba30f624e192bf2b71a29f2a9c82f71..d7f9cbe40b014e3c0eaa40aa6332b07232694c01 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.q_proj/_s.pt
index 503d4434b70e25799bb09b1483f3e928e5a8366a..e43c51c1ce3489f1476a386d41fff7f14bfcfb04 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.v_proj/_0.pt
index ba336c67e865a21fc0a679ef2b10f1a71bb6607e..7c73215b2b22a4b3c282de83e8dc15b1938e06be 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.v_proj/_s.pt
index e2a5cd42f0d0257f9e51db1e7fbb16d0a922ede2..10b6205e85a582b527e001fe4fd750b8c8a2030f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.8.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.fc1/_0.pt
index 1db0a11f3cd51ee38e20872d67925ddba833c4d3..4ba27284402d3da630ae42da4a806b1e89e79c7c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.fc1/_s.pt
index 1c7c74950616a1c4d0495371a1a8cac2962db0fe..a825b43dd319cb479a73a1868db0df7dcfc3cf30 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.fc2/_0.pt
index 7bf832c7c40f4d29b1be0d491710260d56e6044b..893bb6681d1dff94337aaa719590cb0148d35aaa 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.fc2/_s.pt
index eecf544ccda4ee23c9d09c431a68bef7fedc5ba7..a21384693d488503d98496cd146f520af56c9904 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.k_proj/_0.pt
index 060778f0fbdd5b07433541ffb9e94d5176cd0553..4995ee5f0b245a5221668c064852dc9aae5d5e1a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.k_proj/_s.pt
index e90dda0724db3f5489795b5eb242777240f46e84..cd1d310f0f6ff32be944c942acdc4b5d4c4ca38e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.out_proj/_0.pt
index bfd00f965300b0184556a1f5c94240586df0b34a..61a0fe2f8a072f32e10b4148916a17ce10ba9995 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.out_proj/_s.pt
index 2a804846b2f195510e7060c556c146245817e824..7b8368acb93457f9a6c99150d00345df3d95c512 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.q_proj/_0.pt
index 7d1be034794d989df55df7c67401e2f0c337824c..fa23c0c04163ba02e2c9438ab9452b7b3ba50433 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.q_proj/_s.pt
index 81890d118abf514682d966ef82888382cd2ea31a..fdfdc99948ef72c13fbf38f9d117898c1e675e05 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.v_proj/_0.pt
index 4fec1bdc090f16d3e9a857f3216f759eb0e160ae..f40ddc4a430711461abafefe312bdd50aa78c3a0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.v_proj/_s.pt
index a9d643eb9d56dd6a0f4d3d24e360522d9709f94e..08113c65a62255825ef3275bce4cee288c8c7f32 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.layers.9.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.project_in/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.project_in/_0.pt
index 340c7cbf87dcdc7318894e1c56e5c21d82a92111..2915dc695240f6dc8d75efea6b928056aabffe51 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.project_in/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.project_in/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.project_in/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.project_in/_s.pt
index c752f2a574e40f03071a1455b3013320ad64adb1..ddb1740b6ce3db23595cec31ac191b1281decbf5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.project_in/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.project_in/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.project_out/_0.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.project_out/_0.pt
index ced1f6e07f6745ad311a25f279c793ecf5fc3f30..a42897c7ebb7d74c25d4af0de3516b1deb7957d9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.project_out/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.project_out/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.project_out/_s.pt b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.project_out/_s.pt
index 7a968e6538f52bb369dc8f760b2ea388cb885cc8..85764d55d4441921484a191e1f9230db257d41c3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.project_out/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_16/trained/model.decoder.project_out/_s.pt differ