diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/lm_head/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/lm_head/_0.pt
index b26634d31627d1a0044c3f3651ff144ed5eb8f4c..257f78e580e29f2b211849ea7bc8ce9dac122cc9 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/lm_head/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/lm_head/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/lm_head/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/lm_head/_s.pt
index fba223cea433e1767606750c6dc7f6024f468602..73995c5c5d713b6b38a34de5f979e3b2a97b3cea 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/lm_head/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/lm_head/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.fc1/_s.pt
index 1de26a0cabdd4c911bd845ed3770172dc3e57f1c..f8c8ff59e8089a533cb75338c0a33846894521da 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.fc2/_s.pt
index 9b13f1eed690e4bcc5b3caa1b20ea852d0e9e729..810c49da90960b033c967c844da135f8b6811433 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.k_proj/_s.pt
index 5147876fe340014d2e19fcb8b3cbe532d234adbd..8d7c5a82942aad7b2db0039febd2d024058c0c87 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.out_proj/_s.pt
index f6272b63f0c7e23f09aa22fa21097c54f3c8a736..f4733136be79b9433d4ce409b47b05442f557774 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.q_proj/_s.pt
index b583d3094635a7804cfd39b2810adc3c74e32a52..be9c45d2ae68f0eecd85dbb2dc18433e9850b6af 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.v_proj/_s.pt
index ac480f16f2decf9d9133b985a233f1d915f4ca05..9a5b90e6c76c0fb8fb35b31af2409e2d6281bcce 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.0.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.fc1/_s.pt
index 19bb4e2ddfdf5705969712f792421f3278107405..7a235f18eff94136b6bb4c514d12cf8d2a9097ac 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.fc2/_s.pt
index a94ac8f6661ce4f62d2edbe8c386a3055450c96c..9c19695bc4850ddf671f1e14b5f946273c6fb38e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.k_proj/_s.pt
index 95b007658206035f2b324047cd3dfaf118d61bee..a570a3b6041ba835d10e247e11619ce4a13b8c7e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.out_proj/_s.pt
index 46277c77c465fa494b9e6d8b8559c8fd6ef1ad53..f34e34e8d8f4c845b268d0e9c157d4adf6a6aa3e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.q_proj/_s.pt
index dca7deb25cb0211ccdd05c92e7b82c23c6100487..7070fe57bed0b23d4b55e20bf6db8939869d071c 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.v_proj/_s.pt
index 91a29d009c51478745153de0be98ea34c621efc6..45592a929fc55fed74022de4845b089b8a9b74ae 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.1.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.fc1/_s.pt
index 2294e756722fcb28fd4db3d8a9b356167046d9f7..1d3e69ba9f69d3b54fe33355baa6ec5e57783fa4 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.fc2/_s.pt
index ecaa74a94826c2eea8989782563f46ecbf85fe4f..e9f2e2cb0b5454b5fa4520d26c7a691abdb6907f 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.k_proj/_s.pt
index c8e2bb888dcd58803614c82f9e864a9b18eb3e20..3d21185162668ea5d5551b7bc465106d7dbcb786 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.out_proj/_s.pt
index fe186bcc939add6140ac4d997b39fc92048c291e..b3f0341046cdaf5162ca8b88b3264c16dfc06a32 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.q_proj/_s.pt
index 920c9f29fa9ee23c0deebc3c8b95e0c453a1ae8c..c2c19503ef5fe821761768c4d890ab7b7d786cef 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.v_proj/_s.pt
index 0379a355ebdba8be2e60c1d336367a595a091c41..b8e0930b4dda55411ef93317ae99e7d0e3f24633 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.10.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.fc1/_s.pt
index f456df2299d9fa6ef8a7c5b34682c9ce3a19bdab..394385c9abfc2fc700d1483e3b923cd1d5b40c0f 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.fc2/_s.pt
index 92bfd17a01030bb51d8c349c8e523e551251d350..f43a797dd70e16a20966a6df5d8423a62028a5de 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.k_proj/_s.pt
index 103666585618899d4d7ea74b4301bf2f9a695c18..6542f31f6bcb430c27dd99409995f4e0b67ca6d3 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.out_proj/_s.pt
index 3eebfb0b47a946c4b27e9f003aba86bd4d159664..98473849549085cd1752899834b77437ca9c7c58 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.q_proj/_s.pt
index 5c1fb4992cda541bf2f4ac1b927fba1e3a2e0b76..3b6ad9c78e78ef1aff3d153c6a6ae1ac4519f8c4 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.v_proj/_s.pt
index d115237cd6d06b3ff20dc43b210aec59a7e5bc5c..322a32a0528e21a1f5b9339c566fd6abfb776058 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.11.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.fc1/_s.pt
index ac2af8f08892f5575c0c006da41170a834ad470e..92a7b576ffd8325672a479dd65b6aa66c5b9e2d0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.fc2/_s.pt
index 517a7c67ecdfc76623c974212ecdb7d6560673d8..6ace52786af698bb899cdd5eecc6c82c9a795bc7 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.k_proj/_s.pt
index 3bb02d8ad12b30c61516e3dccd2035a85b7c2fa6..54a4da8920336480c773f1cd11a0f02ea4e5e756 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.out_proj/_s.pt
index 41c971e72a9d2b82042a8f4bb8ee6d2cab52da25..d734dd6e12824e9812fad4b693ad1e2bf6de2146 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.q_proj/_s.pt
index cc4b359efcea8b46c22c9786f9b81a9b9f84ad51..7dc3a0e2d2838fd66f226ac79daac6eebdb9fa92 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.v_proj/_s.pt
index f78c43d39b3119244534c326f7ff8ec9ffe1c8fe..5c5d606e6ed9e30871abc349ef4ca4e31957a705 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.12.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.fc1/_s.pt
index a75790de42d06c5e1b714d1a932b23230221756e..126c9129466250736d4e5221432b04b78ee52b22 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.fc2/_s.pt
index 0859ca2062ef37e79f18f1253d5cba7035f9bbdd..ef06473ae98ea114dc3ec758a38efb8d3a84d38c 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.k_proj/_s.pt
index 45a1c9aa75574c814a6a52baca53845d5c46d35d..4f6698171416fc7bd5719eb352c54d7ddc1de836 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.out_proj/_s.pt
index 18b78b2b3785a8dfe9621d26684eec254a5198f5..22ade1776c88550c17a6b8096728ea890e6a90bb 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.q_proj/_s.pt
index ac88e9e420c23f524022564bafded770327107f5..35f53a4b0b2df31eecba63f013c281313637216e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.v_proj/_s.pt
index 6d6be761e91adbfba7b58a30b3902b7326453328..3d06c1bd423309073acb47335c1704bdbd60e72c 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.13.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.fc1/_s.pt
index 66c90124daa062d10345647c9599e57630bbff1b..30883ac60da7cc60e5f5fc567e0be2a43fcbf5da 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.fc2/_s.pt
index e12bba35c1f60c9a6dd0be2b4f95ed72f4a4128b..d0ddc6782fb2906654380f82ed173c9c4db83f71 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.k_proj/_s.pt
index 091a678bf9ba92c14c947bd5ede11e476f97069c..a76cda10a779e544e1cbf3bd66e571513b24f6e9 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.out_proj/_s.pt
index 52805f22789d73e50287c29f237b85ee1719633c..f6bdaf87e84addc8c737e94980f7af62b2493b3b 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.q_proj/_s.pt
index 7b9c4dbdb39a385d6fd0420fa320dba8d1ac06c2..ffc53bd22cd3f99d4178839b088b08bd8edd74ae 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.v_proj/_s.pt
index 1e3ad08d42e78629da20f33a74c368e3d18b9a37..a5e0fe1b6d3c0c16ae03b4644de7818ddb7a25d1 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.14.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.fc1/_s.pt
index 1f2a0d14b7e4edc27ca0e5eb711d3b325840e991..271f46671af7b8b26d93360a7f96699b57c83775 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.fc2/_s.pt
index 2326a318874e93dd8a896487673443a1b27fa610..af4fb87865fbba358cb1738c93859dcdf4e59e01 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.k_proj/_s.pt
index 86a5e6c7ba42ac25a1156b17255488672e84d925..248dd5a5382644dcc0d2a4a6f9b9cfee0f2bc2d1 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.out_proj/_s.pt
index e0e51ad2172acf7530540278afa3f90d24e61066..89f27752da07cc10755e843def4cf363e24c11b4 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.q_proj/_s.pt
index 8b6fdf647962cf043deb8c8ab09650665e75260d..d99113c6650748792eceb206bd38b8bfdf49bd19 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.v_proj/_s.pt
index b70fbb05f24c822248331f58a618d786397c033d..6b5fc16cfa1f3a775cabc95bf594b2f25acd2c6e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.15.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.fc1/_s.pt
index 3acb488aaa82e038446df503d5f90d222198fa2f..12e96e5cc03b255dacb8ee61e1ef64f57f79af4c 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.fc2/_s.pt
index 3624085827bd2681478c02415484f25708b93307..3b7a86279e2d4d27de566fd34a510ef3dce2aa86 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.k_proj/_s.pt
index 201e9f271fd18b30f826f51ed672495bfa732c5c..95d67488ce66b1b0e67a53e431bab551de5b4e22 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.out_proj/_s.pt
index ba698d17c2cf894353e0bc2dab332127a81eda97..dd0be02fcc5abc75022ed8fdde8a954d9d94e5f5 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.q_proj/_s.pt
index 964349599374ee5af7fffa272a4346a013e3c584..0124f7238c7de8c58a307a8069d00f2047235e56 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.v_proj/_s.pt
index 873b4b66eca47f93962c591d29f864d9fc1d42c0..777846ec0710e73c8ec1a60b788abefaec95c76b 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.16.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.fc1/_s.pt
index 7104c63b4d5b661dada0aaad31e3671d23f478ec..b540a80406dcf0aa284f4e774ed60ae0c1049ebe 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.fc2/_s.pt
index 214ce679d6fd8df573017dd1a36b4f8c33fc7717..03838591a93bc7d3f097325b9cab5acd5ca8a568 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.k_proj/_s.pt
index cd3a030a246fd92fd076ab032c5fb2fb936012b1..5a97cf6cfacdc0502f0afdbef90f5af2f5ba1bcc 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.out_proj/_s.pt
index 069179566e2c81cefdaf6ed794d21dc83e93cece..60c79988532f7772d1b0eea55283d45925444394 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.q_proj/_s.pt
index f68531c846aa79473e3a7b2b5eb4e24a70f8e331..4c550cba76f0de29c24a4a27c21fefb6711dd428 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.v_proj/_s.pt
index 3be2845a703efaee41abf956a0e3b0cc63f7e4a7..519f3775aec1751d600a8edaa29ea7ef0aa501ae 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.17.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.fc1/_s.pt
index 7953aa798b83574b5f7882c28de5780d7d2cbfeb..fbb2998bfaf9852d7a4add51ca2fad6cfceffe9a 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.fc2/_s.pt
index 2fad01d7640b08c217b4911d47579294430b3b6e..08104b6a3481ad026d1f1d52b3353bd5d0e98541 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.k_proj/_s.pt
index 2ea1aa36751f76fdd5b36a818ae702fbc1d73979..f93e0f8a5d613b58ed9229f3201c73c46b5d7e16 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.out_proj/_s.pt
index 4deccf513cbe0007b79785860cc6c12fc4beff20..014482b841a61973322793f4e317ac2b58a27e12 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.q_proj/_s.pt
index 5fa2eaf56e8d837af6b3a2bfdba9700fb2f9d71b..51ac192323d8e5286489298adc321086a4a51dd3 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.v_proj/_s.pt
index fdc3a351f613c769d73fc6591de87a27ffc2dc8e..d933a5ce7aa48dc4fa0a9d0d27de5c7bf94b4e2c 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.18.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.fc1/_s.pt
index 659e1b89c07f929b02e732be335053ff16024eaa..cd34425a1cc4b989c38232e11981bbb7c0ffed95 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.fc2/_s.pt
index 0ccdd7354e378abca2ecb68539aab5cdf8b2005e..4cb8620ab41a9f446d97d983eef2923f62951b3e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.k_proj/_s.pt
index c5be9675c85254ab100b3cd2f8e468bd5f8b093e..85a7089832a5feb35db4b0d4da06ef6c8087ea24 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.out_proj/_s.pt
index 0c2180211e77bf4fc3690ecb06d590db82744026..dfd0037a4eadcdb02e958ddb0540a8ecf73c652f 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.q_proj/_s.pt
index 07ba2180deede085f6bf9c11904a2b047ee38a01..3fac32d0c8737575e142897e4b2bac544e99f4c8 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.v_proj/_s.pt
index 154c1693681ebfd4ffdce5f40fa38e4e3b2b217a..21fab432bcf155bc9cec11fa332f1a1f312b8b3e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.19.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.fc1/_s.pt
index 39a1073b8d9582d3adfa565def2a46a11bc3fc13..b118ed4c65fb5f2446cb3cdfd43c3638d386863a 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.fc2/_s.pt
index 81befb07df5f3e0daae49156bacbdbbb5b0dbc2c..ea8e25844d19102f748c6ed90e3910f3761eeed5 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.k_proj/_s.pt
index 263371376a4d4b2aa07ae0ba50c2bcabfc5aef49..8a79324bb9506923a824d62798dfa0fb88965cc1 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.out_proj/_s.pt
index 726e2c8726cb6343a807767aac4e9c6aaf4ef60f..c70420b5e785a33734d400095e91ff90c754725e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.q_proj/_s.pt
index 5dc9cfc9392163ca4888163d1b3f8eb970ef8e07..65dbce4e8cd873a38c993c3f8fed45dbf60cca82 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.v_proj/_s.pt
index 29d78834374d54bac449a6da64af3cc7c6705af3..449394fb980278c69f90b3c49c8df6e25ce6ea21 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.2.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.fc1/_s.pt
index a1ef4907f04012ab2376a129dedcd666c108d121..b591b16fcbeb858903b9f992be8faeb57e84d5cc 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.fc2/_s.pt
index 8a56ff81ce95888cf2d40c643fcf07ae6cf7c12f..b3e024a6b60f0fc2b15cc9cae2132c3f92c4f27c 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.k_proj/_s.pt
index 2a2024ae469d9fee47de5e108bdc62d8367aa8f5..0cef4453246e68e6129937e124a86f87a3e24d2f 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.out_proj/_s.pt
index 95fd4fb0406242369686ab6a6a32260fb4de1725..14311e0dee58c5f1f277f835715b82936d41d2b9 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.q_proj/_s.pt
index b466bb106b3f21419262a7985e1b1556aaea1ec1..6e0ae56c493f0a595d2f1f3ebd2551d9e85ef4ce 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.v_proj/_s.pt
index 876cf04fe0c6850064e38f05dc31f703499ab5bb..ef6063fd3d68acfec9716cf12256f92557cc2aaa 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.20.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.fc1/_s.pt
index 463012b77247f5ba7e516f86d67b74d40e9fbb6d..c237350b825000c4c063ba471a229eb34b42c0e7 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.fc2/_s.pt
index e2a4f4b220a528953fb77705164640842d69a404..ccbbfab0adc9113aecff13786ce26ad1525c66c0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.k_proj/_s.pt
index 8996dfa59976547e77a3db12520a6cadaf3a8fb9..9ca4883dd0747f7b76cff22a2e8b0f1871a252d4 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.out_proj/_s.pt
index 47df25528b75e506fdfb874abf1ec2810080248c..d0f4b16721f352388ae602cd1df8dc82e8eab417 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.q_proj/_s.pt
index 2786a7644f6a674bb404d26fc60ed888809421af..c19477c7e3d5bb2e39931969953cb5a86d0f7e80 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.v_proj/_s.pt
index 264d0476c7a9bf9076b7bfe078f1d802a43c8ba4..b8429e018c419a9fc5543b353e343927c39133e2 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.21.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.fc1/_s.pt
index 8bcdff37c1cf4cdfb80d3890c5a88fdf6ddbda1c..d96cdb49c5d598cffb6b3e47be59fdc81b1cce92 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.fc2/_s.pt
index 675f0949017d54065df3022bf582ade42aa9b955..43623cbd97d75edef2dae14b5a84fd23755502a7 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.k_proj/_s.pt
index 408f7c117586f5b6ac005b8024e2778311c79b61..bf7e04cc8b323b638fcf353b3a463df86c9299b6 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.out_proj/_s.pt
index 457d8dde2a51a146ed75c7c91c79fa63b432181e..9fd6b63f94b01838d4fa39793ab3a4cd29909cf4 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.q_proj/_s.pt
index 8e562ae4cf5ea0c797ece44d8f786aaf1645927e..e05fbb91f929d0cf5cccdabf01dc87b030260668 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.v_proj/_s.pt
index 147cca1260f5fb2553784fb3269d7cb08fa55160..856d6f97f41d7c7d867a31816439b61e985069f6 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.22.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.fc1/_s.pt
index 0076bb693eb1224971232a983cd1b7531578441c..4c7783a0b0602f1a2d17eda511288aff586377a5 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.fc2/_s.pt
index cb8338aae85810af79ab98cd42ef9124782ce1d5..e97c6681c39829ff277acfa1e26e677321605806 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.k_proj/_s.pt
index 2c0111629effd76a4939b14b5ed2426c73ee952f..0213b6eebffe5fffa92e8e1fd8ed5c10f861fffe 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.out_proj/_s.pt
index bd9cd32bfe92a28b161badaffb89e42621b50098..e5d48a492e1e89b4eea906d980b023fbf074d865 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.q_proj/_s.pt
index c4909c7cd567c2ecf3aeece929dfb7cc800c4838..ee3c196ef304fbe1749772f56e89baf6f63a73b0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.v_proj/_s.pt
index 53df7368e515fc52019a204a9045a3d79191896a..175d4e40303bc7ce15b7348d400d7c03d2f8dc99 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.23.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.fc1/_s.pt
index ed9b4523fab103080248be63dc67d7c870d8cd4d..7f4cecd36d5075a78ca9dae831f81345e20655a4 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.fc2/_s.pt
index 9362f4f0d2c457ef303d4b8c9aa6b196b6764a15..c90fd53145eef9ea6e73be569479bac51247d61f 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.k_proj/_s.pt
index 445ca91c06dc44c013501489c2a4b91cc94e5bd2..3a9477ecd1903597d469d7982186bfa1f84c5603 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.out_proj/_s.pt
index 6f64d8755b65a7cc381351455ec0e9710e9d65dc..a815acf5671b6a7be88e6bece1696bfd18557761 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.q_proj/_s.pt
index 2d7f03b2267ef2380ff9372e3b512e61049b185a..b87d958e6c9194073bc81c175063084781daf030 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.v_proj/_s.pt
index 628f0a0834f765a90ef60b6ad76b92320f2b5727..5d9a06b312d3e8a87f0b81d0f90410c4f6bc9b95 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.3.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.fc1/_s.pt
index d870826a04b2dcbcbd9d7faa84fd21b917b822f9..9397f83fba007d0793a9088c75e585b933011640 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.fc2/_s.pt
index 539914d84b1f9f549ce2ef9284d2109561016af6..56275b3d1acfecd49a1ba075012306c8fb113ce6 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.k_proj/_s.pt
index d589b3654b102a425f749c581575ca4a6b79e545..7866e17696fdecc076035b31a47735405728fef4 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.out_proj/_s.pt
index dc82bbdc24e52d78b5aff44318ebc4cc1bcdd80e..b9230be204f108b0dc1ea1d306592ef05f871657 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.q_proj/_s.pt
index 7b08f8c2cbe5743550c08aed2e28458e5adf724e..305db2a8844b5aa3e1b423d525a66daf71221fb5 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.v_proj/_s.pt
index 203fb765aae0bf0bc201fdfaed13ea554cd1a931..31cb8c844ab18f6c4ba821f62887d124f0ddad87 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.4.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.fc1/_s.pt
index 1ad85e117aeca1b0e29a5f40b3669567a1d8a9e8..ee8708a2f878ae260d4ba44479f94e06ad463453 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.fc2/_s.pt
index d8f7e376af6c0a3bebf14a4c2194987d096d57d5..5b9b780142b0e97169270f38f6bd207f18b420ed 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.k_proj/_s.pt
index 743720fcca6b152c21d2c8d3c053a7994b592821..7389ff80e7af6d2a0f412c64ddf1a29352e63334 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.out_proj/_s.pt
index a5333ae0fb72608be7fadc19eba87a5abae68e4c..d221d862cd4d28534de5317a23cfb90cc7813e94 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.q_proj/_s.pt
index 27dadcd85384b15ff3049f9237b8e4a67dd988a4..a69675e699ee776f544ddeccf7c48aa830236319 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.v_proj/_s.pt
index bdf1068d2c1d0ac29593da9c5f403efe0a25378d..b3f0063eab96f8639491cfbad352728001dd005a 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.5.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.fc1/_s.pt
index 489ea743d7fbf582dfc4aab635d8f8e6fcf44c37..08630f259fc545a560a6f7b6655f91d7efa463c4 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.fc2/_s.pt
index 3bf00b42f21e1c60b2af0986398ad28e0243f683..ae5c34315aaf16581786a6c00d07db115ca35ab2 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.k_proj/_s.pt
index 1f74b0703af4310a3f2ad6a5ae641fda0ace9e18..0590dd86c514a481280517ecb34118b8403fc513 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.out_proj/_s.pt
index b32bb44b428d4da784e92237ca913fb0a1d93ef2..8f8c35ed99d933d77f049bd4b8f1ddb6dcc51bc1 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.q_proj/_s.pt
index f9b6eadb232301b5f4ac5adb1d755f5e5d14a85f..d815b3c94a15afc4c53f1986eb9e07f8f9e61c81 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.v_proj/_s.pt
index 4209a5036fd3b76e5d8fa578529768c9b86312d6..b45e8fbc8b7668b1ebd7f284e0be341233511761 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.6.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.fc1/_s.pt
index b07354e14b41a4efe0443894d1fe3bc1f57734bc..c6cd55b34abcd14c703a9b3d8b30fa7755bd4485 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.fc2/_s.pt
index e8abc27a19271590131e55fc6550fecd288c6d5c..d2fcb473bebebd768dcdb1bcbc9d662ce6a141dc 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.k_proj/_s.pt
index 4755fd2ce3f3d6d9f41e34beb6cb3b026b6fdfe6..53ad67d56dc158d668b415234f84800a175f2939 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.out_proj/_s.pt
index 6cea7fb5034b52f7d16acfaa7408bfece7be3329..b8db0a9e02c8bef5777db7b865933e014aff6443 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.q_proj/_s.pt
index 36201e291f2b2c9aff2516b7d43ea6173c7ba232..d2312e09ba8c470424583341c247e78ad4e33586 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.v_proj/_s.pt
index 1962b9a0b3c9b2d91dd8444a945650b07cf6f0a8..b57da4683749122d86bd61beeca1d7d9911e920e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.7.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.fc1/_s.pt
index 0f7a2f72f1d388ef33c47a7307c6e5ad2310d696..91ac927556e8b707d055e5828128dad6563a164d 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.fc2/_s.pt
index 395925944871a38f7e0b931351bb39622fafd1fb..6c5a5dc83cc81af3dfa0bceffae407dde26eef9e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.k_proj/_s.pt
index 415555ff67516e9a0404d10ef79cdd6bc2c0ed76..2607c5a1bcef8f3004895a91f6a108f4602f306b 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.out_proj/_s.pt
index c858bd25af2be7af81f360de5191bbbe5a8200b1..86f654e752f5ca3611aef3e8de08c62352d11624 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.q_proj/_s.pt
index 8a377e911c9b950ac2403c5bf2fcf56e8be5aedf..158031423eb0aa7620ab7f1f87a80bad3245d856 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.v_proj/_s.pt
index ef920f341781df9a92f27d26091210dd696f3cf4..e4d25cab69b8c9fd3be782652799fd65a0579b48 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.8.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.fc1/_s.pt
index 8466a0b7f553d247fe2f150c18dfa5a599657d7f..9c264cb4cd98f0a67602e45dba0c603096a35065 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.fc2/_s.pt
index d781b706dc91dcc01c7cbb787cb02eb1d17f5a13..8ac98431181922fbc733c8874dfd957c21837776 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.k_proj/_s.pt
index 317a0ff7100194ede82e7a2891211a827012e989..7074223b8597c0239964966b16107c481954e15e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.out_proj/_s.pt
index 4a22cef1f490ebe378d5f90440f7f1ac6f78d29f..43d31358036c44c7808aadfb0eaa8886d1675f4e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.q_proj/_s.pt
index a04fc32a7a9eb3e4a6f9e6c97a81e8c660870228..1a25871d3d2cd72d8e34004f2f0625c0678f197a 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.v_proj/_s.pt
index 69a48ca4d95ec5de5d4aeb8542b2baa1accc1ff9..8683085d3b9ef095ca12da4459496db9a5069208 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.layers.9.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.project_in/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.project_in/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.project_in/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.project_in/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.project_in/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.project_in/_s.pt
index bc26c63b1a8df8abd5e6e9a571d607c5cf15c570..75ed45cc20834bec96098b2cb6c9ad7cd4f98b7a 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.project_in/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.project_in/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.project_out/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.project_out/_0.pt
index 7dafd7dbb653cbd2daae7b5c32e1db4207127f70..6f3ba354ec876f170cb0943973ce2085a27e0750 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.project_out/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.project_out/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.project_out/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.project_out/_s.pt
index 9ea1606f693d8b94381c2ae311a30f46f63c0a72..f6e4cc1580608b2f6a9c3985832e66a41bbbc528 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.project_out/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/init/model.decoder.project_out/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/lm_head/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/lm_head/_0.pt
index 68e127c08b44ab018819373bc4a3334c938e7216..0a0eb8f608261a09e53bd9811d82ff6d90e4ae4d 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/lm_head/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/lm_head/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/lm_head/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/lm_head/_s.pt
index d2542c83c9635051c850dc81d316f00a6ee6fc0d..f121a51394cccaeb15232dd45a6bf21318d26e2f 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/lm_head/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/lm_head/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.fc1/_0.pt
index 054ab903d9bff89ec344d70cd894ec374ec6f352..11789e6adc1c0c8ae86d2a7596045077ac018e93 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.fc1/_s.pt
index 1702877a97cb2331405d26a56e7e4a1d4275f55e..9897b5901b3d42320fce5ca0befc4da33f6e9d3f 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.fc2/_0.pt
index 1266e7219f2a86a9aa89ab98cefc062def1ecaf6..f19fc82c00031af2c90520b68d07a99adc72dad6 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.fc2/_s.pt
index 78bdab21d7e7166c6416edfc331f6e36bc9c37cf..3dc5de9c095eaa77942b88d6e76df63bac235a12 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.k_proj/_0.pt
index 4940d99791108b5a3dfac936d84e8dbc45bc1f95..11dc2b88e02647443d470db6904fb1b784363a84 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.k_proj/_s.pt
index 96baa8d9dbe426ac77843b062240ca18df31fdde..0b4b22ebe16af75b85580339ad50561c1a118f75 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.out_proj/_0.pt
index 2bccc956d7f60f41d5e09321c926c39eacaa6d14..c6d4f8a54c211440efb92ecccb5762b877b7a614 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.out_proj/_s.pt
index 0cd00d04e2fa80bd026b68a2c0136f11ceaf1ed8..498e602d05b1b197d0dff7b0dff9cea4b2f5ee4b 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.q_proj/_0.pt
index 2a4bdcf07b7d5d08cddaae225a95548fe660359e..8cc81450cb99e88e77b8d18a66925514c183ec93 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.q_proj/_s.pt
index e58c3b8448f82705d29c913e0161f9c8b9d09039..dfdabdd310d8e265463c680d3d8b287980e9f797 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.v_proj/_0.pt
index 36414f56ac64013d603a1e8678e8f9adf65477f3..99926a758a4730ad11eaac58d861fd17be3ee5a5 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.v_proj/_s.pt
index 39e175c5f18fedd561500fbcce62d38ebc7d90d2..54ffab7ff7723a306f4ae3502da8f91fe55942d4 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.0.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.fc1/_0.pt
index 8876c1b711317ffb38c361f26971947c8cfafb1a..bb2dbcae6f94fdc82fb9bafa775e615f36fd7cbc 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.fc1/_s.pt
index ee936ea447b26441d720d6d399afea14d041cdf6..c4ea14f04743bff2f9ef8a569c9bb63159cf249b 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.fc2/_0.pt
index a6172a7db600781e858db865fbc9a4be4ced7878..2138c5759d7159a3794089bf5fcb75f9b719a0b5 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.fc2/_s.pt
index 3dc7d4919c6dc7cc299d89fb5a58115e36430bab..3401cba8c19e254303890f99db82749e8a22dab0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.k_proj/_0.pt
index 2d9d900c5e767217568051c7ba5701473090d1b2..8bb770824204163fcae54476204540893f9d059c 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.k_proj/_s.pt
index 1697742d5ed60c7f54ba68a62459089380718cdb..6f11289cf9150482d769fbd3f708f0b38ee08545 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.out_proj/_0.pt
index c0245882b2dbaf8de409d36af744e17b6a9e7b54..a3604c6c7721ff0fd5ca39becb84cee94eb5b8da 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.out_proj/_s.pt
index 52ff9c70e5f1af6784bb17a5510d979af91e005a..26744904fb9b88efb21d2d1ab4e1e728783bde82 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.q_proj/_0.pt
index 81a0fae6f3d08cd1bb53f7d4fbaa17e22a38b2ed..140fd189388bf9fe6e9da6c2f000e90e7b451aa5 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.q_proj/_s.pt
index 71c5fb6e3836506f80fbad64785ffdfa036254d2..5a50e155abe3eedff4c8edbd0918cc78ae744c08 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.v_proj/_0.pt
index 1c84c0ead48840febda86c67acf1ee8475b8ea14..7211e510015a68e4fd780aeffc3cfa34241ae724 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.v_proj/_s.pt
index 49bfacefd5e909bbb5398e158a7d62f9f3aeef41..4180e5856eeb6d52c978f5cf686facd4f5b51725 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.1.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.fc1/_0.pt
index 57ea1c2bb0551d0ebb9a94eb3793cfe38c82db24..787636f5bc06d3975be35cbaf53d036b688a556b 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.fc1/_s.pt
index f38580c62cff8b8b2e3feffd667907e54986da9e..afb171068f05b65f34ba1459235e58b5a22a2972 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.fc2/_0.pt
index 21b443687dfcc5c4134b391fcdaebc008b068764..bca794a4318548c56f65e7499b0850f57b2cd65c 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.fc2/_s.pt
index f4ba896b8d58fd1b148ed8dd8f3c2f73dd6d9a5b..84fd609373b95e632a11b2985bfef7173cc09b95 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.k_proj/_0.pt
index 6b4ecab3e149fc66c6ac79b743b0ac31a5eaa740..a86efe8801a21eaa51cd651c502a4d07d58289f6 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.k_proj/_s.pt
index bbe33226617d44dfa691bafdf8fde539dc27f393..865479986702d510483c69cc369af102c5ae5be8 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.out_proj/_0.pt
index b6c1b1b625e2624441bdd27f1c626aea59a4385a..75813506b2ecfb3d4f444c7a71c32e35c6eb9d46 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.out_proj/_s.pt
index c1f127ef95e6dffac0b41e521f90d08973c242cd..6d516ba9b6b209297737476a649cfc28e37d9921 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.q_proj/_0.pt
index 6880a6f05f4fac3724c25dcb048756a6435c4c66..820420fce53f5de487df9be89cb689f840cf25d7 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.q_proj/_s.pt
index bb00e6c6a4e044351f3bd33ea6181fea23d071af..f2ea9e6b25a910953dba055f09ecbf85ce5dae44 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.v_proj/_0.pt
index 779cb328e575e647205d1a916706cd930c92652a..585a716335f2eb9d0207781bf9e45f2cd3416373 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.v_proj/_s.pt
index 2bae76acc1dd391a35774b3e0a6723c018b0bf44..e0a67af5f4d758aaba04e789ed69200aec072cf4 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.10.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.fc1/_0.pt
index a3d0eca9b14a9df691e645ae7d1191439db771c2..327f466ccc2816e27d4a1fc9211fbeb0ca7c38fb 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.fc1/_s.pt
index 61e4c9379f487d9b32772a64621788c943d9bcba..2e3196ed95e79bd82a5e087e6c01d8d7d473b432 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.fc2/_0.pt
index 0817c373aec83fd18f7211bf5b14198847d684a8..aa322bf84015ea905baac5cff77c20e373c6eb48 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.fc2/_s.pt
index 52eb3cd885b7647390856dfdc1c6e13c021ec425..273eef081658a8a5059b9d64d12926e4302f81be 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.k_proj/_0.pt
index 76bae0faa802cecd3acc3930b21830a73feb4459..14a9ea1dc41735258c55c1133664b1ead96cc253 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.k_proj/_s.pt
index dab3e4c3e1ab2ed46813c88a4e6e2a7addb5e68e..e92ab3a4f433d666fece973b95cd4437c3d45889 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.out_proj/_0.pt
index f56393eced2942c767e49c898cd6c85f085ce79b..493eaaa820f56377db591c596be65f990cf33208 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.out_proj/_s.pt
index 908d192b24e9c2e6edc3c13f8131a6f89a0d7a15..1c6bc046fd22e625f3b4a8e0f62177cc1e3b77ae 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.q_proj/_0.pt
index 771384af0c55e375f8cb2e5c880469e16ae43c7d..f54d4aee614cc7701668bc512846ee86ebf177f8 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.q_proj/_s.pt
index f61ac6fbe845e3a26b2f21c006f098d41be0d604..3b57f3487b00233e53be5dc462a54b81c7274eb3 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.v_proj/_0.pt
index 7d870dfdaef6eb3367d910ec84a760a444aa85fe..1eedcfce1586970f3835e9b0dfbb4dfa8632ab51 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.v_proj/_s.pt
index 772ab813faed447b32616aa6fb1248fe4d67e9e9..5e94d5544c859a53a9191cc1e9e5e0c381771c12 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.11.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.fc1/_0.pt
index c4efe4447b7435749ef968b93959c9e2f1c4a5e8..3162cf45f7cda22bf9920398e879a9d67e46221f 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.fc1/_s.pt
index bb61951e794569e4708633617c5cb606bcacf8af..2a56bc3e28e0b74c4b64de5efd8f9631811f35d0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.fc2/_0.pt
index d112e11081e421fe98643d516a9969ea7b29ac5d..8f869f911d8c30fb3dabad9a990407749d31b6a4 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.fc2/_s.pt
index 5a5b645b5f325c3b1b30450f0540f27db3926c47..0250cf2166ddd093ff044fbd60b8aad5c8a82f9a 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.k_proj/_0.pt
index f75b547024fa8d7eafb737c7c8c32bfa8d943552..ac1ee12146b780eb4e4f18acc9aa7061c78fef87 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.k_proj/_s.pt
index 672cd53fab22a419714e44a3ada69c575280dbd3..22fe6509fc6f013fd3ecb24b7e0d4304314dc244 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.out_proj/_0.pt
index 836bbe676bfbfab728b1fa3a00350bcd76bffeb1..83708a315414806ccaf28f907014387ff551a2dc 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.out_proj/_s.pt
index ae6e5c8d328406015a3097227378252f020549ab..8c2a38dca97c1b4b40c6813018666645262943c5 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.q_proj/_0.pt
index 82748459e76092dd6f50dda7bc203752a300fa25..863457424915e0d4fd66e0aa5e848dad29c4dbae 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.q_proj/_s.pt
index 21bce9c8fe694340ed3aa87261764b66ae6a9111..ca3672d0eb1ce7d6ada2d656af230f6c0a91b1db 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.v_proj/_0.pt
index 4f0118731f6e8d60245c8375b1207c0f1edcca68..16a087bb011d9fcda8b14aef1ed58254f29a31e6 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.v_proj/_s.pt
index 191cecd13cf1c266daad4797c2ad118d2ccfde05..c26177a9c36a2f9ee4fd8f23a90c1bc05abdcac8 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.12.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.fc1/_0.pt
index e576add735ede50c28d31a89a0a25fa61f8e946d..48d72deb59dd3539614199136a6b1bff8bfc046c 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.fc1/_s.pt
index bb798cfe7f108d2b110069a7ab8f2edefa64ccb6..942a3b7d04f7c935451bfef56d83c9f9aa8d752d 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.fc2/_0.pt
index f0023206756f4cd3c18fc5648d652ef95b85034a..4939d895d32d99013e80c19e84803d1906cb8fa7 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.fc2/_s.pt
index 6d6a169ff9f8b786904fbc67b1816076aa5f9614..3f0de34f49211f543f9ed7cca28cb5b9ba1dfe5a 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.k_proj/_0.pt
index 82f75f596d5777ab738470fbab78450cfddaabcd..0228daf0ec8399124f61eb6fb03d708b65c6ed5f 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.k_proj/_s.pt
index 4c9d78f064cd84dc10d739bd3361ba8654db040d..b1153ebc6e9dcd844f31aafa0007a21280ed5a7c 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.out_proj/_0.pt
index 81af95729ad453dbfff4b15fbf2bc2d8d4657be5..85343318e81a1406b953cc0e640148a5ce77eccc 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.out_proj/_s.pt
index 1813ad07feeb714c31455394c0dc783b61b6ce17..d2c301a38c60df75d613177cb41ed18508dec042 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.q_proj/_0.pt
index 98524c16cf74a7566936033a724df5ace82c5395..14e58e0ad28792558e761a28d711874e8c19af93 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.q_proj/_s.pt
index d3385b0aae9cb85ad65f752496a8aaac42d8c59b..5ebc06fbe45ea213d99c7743dbbc132149a603f8 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.v_proj/_0.pt
index ae361b8d9aefbe9a425c7dcc3221832f4e0bd72d..31c35aa75de0b00a372e1bec7e77001b4eee2da0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.v_proj/_s.pt
index 5039058c37c520833aa90fbb4264bfafa5e6157b..dc91669a340a4938ebb5e9dad6cc1e9230e1ddc0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.13.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.fc1/_0.pt
index 3dd6168b4db0e8e9cf694ab1a80dad169f540675..df460bc31e6a053d4b6407e0b79fd66547e0ffdd 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.fc1/_s.pt
index 07a810f2ee19ece90badf1ade6c85992218384a3..2200e8d166aa0ddb2eae848550f6d282a3ef7c7f 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.fc2/_0.pt
index bec7b6e2040c10a78da2a7cd3c8e5e827e61421a..46cacd389b772b2afcd45953c6c09809c336741a 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.fc2/_s.pt
index 87aa5b7efc292795092b049bd7281819f7790056..991c8f0b7a4ea35eac5bcf51f5d3c2756d238d6a 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.k_proj/_0.pt
index 4c033455d62f64376c485b2bb28a656bf5a17c59..612483f77c001b8aa8825ecdcd68d156977b7633 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.k_proj/_s.pt
index 93d060b78c0fbbaf99b79e668ffaac94fb109d39..4e2109e38d4e47c5551d951db8a845de8b7ea1b2 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.out_proj/_0.pt
index f4805ec61010891d58500884bcbc68f3632d103b..3610d1ae7d0a5d5ac92f0097cb01d45ed61b89f4 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.out_proj/_s.pt
index 354ae2a7a14fbbf26d77114921937b05f418f7e0..809fc2b425e25223055ed81a55dff1ce48d5d51e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.q_proj/_0.pt
index ba17abd45bdda127e638b52a10428601f0d17924..9173f11daf6e393feb6c7472b04b887baadde899 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.q_proj/_s.pt
index 8453acdc17c22ae0a3820c4fea5634f2fd966968..c6f80e94c910de10aa5abdc4523f3305ed0b45a4 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.v_proj/_0.pt
index 6a90951215f7fb3a7f8ee97f3cec0cc6dcf666ec..5fcb79cb2f1c41d2db5f223cd0ed8e18aa78c66a 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.v_proj/_s.pt
index 51a73925a3cc94b53422b2ab6e2691d9056c244f..1571d5ff35e137f29b84fb1175c1b50e06cce18e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.14.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.fc1/_0.pt
index 430b898177d88ab6b465a3633028b40977b56e9f..2c424d579cad7241e2a1b5579fcd2e3aac7f351f 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.fc1/_s.pt
index ba4f132db781bcd44ee9fd5a37b9f51797d5cf87..784f07ae2a06e72d4f59c848e865214e2ed3896e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.fc2/_0.pt
index 6a9d1e725d39f3743db335e46b57589408ed0dbb..a85a6486f724bab4689ed0f709553d8da17b66e0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.fc2/_s.pt
index add81be6ac6d3779a740bd21fdeeb15ee499f7b5..bce71da3c7b1514842c41e794d510897172ff38e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.k_proj/_0.pt
index df77df225bed9602bf6bd9e03eb3e75d8bdf49fd..a25ca8e3f5d840f50bbcf3739e6fc7f892062750 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.k_proj/_s.pt
index c5117b530f61d7e5885ab0fa65f9431456307d57..70295d42970477798e30e8b72b4498324daf2476 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.out_proj/_0.pt
index b216dc3a3cb7e62cb6ae61e64c0a6a60c8d9b7b0..f89be9236fd3c40dfcb1b42976c33d4767d12578 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.out_proj/_s.pt
index 452e069748d9d7816c735984732a6773beb0d8a6..4658d161aaeea598b68dbcc98de44c73011717de 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.q_proj/_0.pt
index e90f5fcc347692cab3a31d072d03763cf6b95c25..802a445b7ed53668748b167c7a7a2c3cc0b1a7b2 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.q_proj/_s.pt
index 947f87bf08921891220d2ca108d592cafec49eee..06a4bf9320d50d75c75af87da25f7c2198cf5c29 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.v_proj/_0.pt
index d6336b0d23210d9fda0c87b7fe461410ca593829..a494bcccb57424533ab9c7b4e659da2ae1382406 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.v_proj/_s.pt
index de0694d537537694a75d8dcf5ac8e1e18282569d..a2fdcb6075653e15a13f5de3c6e3ca2a8604b21b 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.15.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.fc1/_0.pt
index 2ef31f3340f9c7c2315d4bd2d52555ac46dc7ee9..10b6d2734b9077a79af782be5284ca7ef1cb7f53 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.fc1/_s.pt
index b8374e545a7abb0ae43a1cee8ad1e54ff276f533..17483ae23cdd341fea0b61494839814746a8ca09 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.fc2/_0.pt
index 89b2f6e3266a12982efab966d0d3bf336c4b9395..e722abe59ab99aec08997ac6a449643cb8d96b0e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.fc2/_s.pt
index ebf563f1b4d1859ea3ff4b509e4dffabe45d4719..ed01a0ae391a63543b4d68caebf6683441d15177 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.k_proj/_0.pt
index 21da9fa9d73546526c62f367c8bfb994d69b9979..796ab5d254b059e4b92f239c239df393cb93797c 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.k_proj/_s.pt
index 29ad2c98a8973c93145fc3e2996c2e5ef0f0d06c..7f5e0700522b9e0919a3320ea3cb893a24df3274 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.out_proj/_0.pt
index 6b8173251eeb77a5a2f3aec9b1e44458abaef148..8f29a29a26d93e3168f36df699ef0e1e9a23ffa0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.out_proj/_s.pt
index 9d3a7c783aca22352e6747682f3d5d2a59e06ba9..54b3b1b991b03a0d335fd81609de5b4f497ffaf1 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.q_proj/_0.pt
index 362a1da0ebb7e78b547724d6da239b5afcb67324..d5383ac467d089f4103f4f6c4f878dd4e5bb8c1c 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.q_proj/_s.pt
index 29f7159b4028db71f430c7ac845a4ed0b9fd1f44..59ca4056a1304fe741bbf04bebe2c583660f5838 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.v_proj/_0.pt
index 762677a184bbe7ef4a7e0a3b39c3132a1907b91c..034ba9f22de7d3d87fd3844773b2516752f0655e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.v_proj/_s.pt
index 50e7535a4d4a634f3f2160da6ff893943a9489bb..818a70780f6909b25bdf0f3f0615ad7be1d2d2a6 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.16.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.fc1/_0.pt
index a50e2b43b812d46e7ced0aa4be1933aaa07fa17d..74a56094894cd8dc7e6976eefe92cb8b1ad4f3b3 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.fc1/_s.pt
index 89242b4b5449f0f232b1b57b0351c5fa146a9aef..54021b72948b9a6884f3be5069e5256cd2a970a2 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.fc2/_0.pt
index 350167e0707e728e79935541634d6cd5e27f741d..eb7f523264bd66be8ed7632a3492b07fa80251de 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.fc2/_s.pt
index e1b68190c3e7107579287ea270afc8eacae0f443..2c288524de32df217a161c0bb5548b088a89b5e8 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.k_proj/_0.pt
index 21bb36815e8f1ca1ce5981866ca50fdbb06edfad..9f25dff712e4bf12826e332b1b3f06e945be46ac 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.k_proj/_s.pt
index 9f2dc1026e47bbb428a9beb03b268afec65c61c6..66748d4525f89ce8cdb25a3e6426ab76e519042d 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.out_proj/_0.pt
index 732bd35deaa5046c2e28c24de3bc4d410818aa35..d0515972bdb2eb25d0a6be292c1e3f4d87847d50 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.out_proj/_s.pt
index b530aa4b718078197aecd36f6eb05d5afac2e5f8..624d4d2a1b0a6500bd203241d5ee99880a654770 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.q_proj/_0.pt
index 357de339d0833b76aed2217193a6be0ae705ab19..805a1b04579ac202b26e1669673693e61b84a4dd 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.q_proj/_s.pt
index 9ef39263ff25e562cfd6e8c262dba5b7dc6d0481..2b0dc428c27e2ea6d447f7c4a4516a2b9bb0de5a 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.v_proj/_0.pt
index a5df00bb323c37991a8fe268fb4e4f3d37537560..12549b041a172c09fb753db363a24d70fa57f8a3 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.v_proj/_s.pt
index e06f0fa5047ab5c9878274c7942688c270975d91..7aed5f81d3098e4a1475927e0c85843e2bcd79bd 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.17.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.fc1/_0.pt
index de183e8459dd8c3ca93e73c1789b5e5de7b8a804..44aa5de7bc9802f33f0e7a6d7a988b4291bc3fa3 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.fc1/_s.pt
index a2e74db393ef161fa4a1804ddafa4aaa1323cca5..881b62fca4da7dd67ab1f057327a04953a0efd04 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.fc2/_0.pt
index 5069ae7f2cc4fc4e953bc1680701769281938823..21d7dd6959c033ab7b6d04f462de316cc74309b9 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.fc2/_s.pt
index 6fadd113aad7dedc2a01ca9e95446c11e36272a9..6ef0192f4fe926016035590d595e5cabcfcab807 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.k_proj/_0.pt
index b455c4b17e38edf8c976da4bda642ec908264c81..aa06904d6def1a3d5933ca0d6012aaab79b9806e 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.k_proj/_s.pt
index d3156f49aad10c42ef5fbb59debe92295911a681..74d0a8e7f448d77dcb8b56a8a2f27980c13afae5 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.out_proj/_0.pt
index d6421402d675f74ef5b1c5bbce02b9e1653cdd54..0faf1751385c13f2a4e624364252309dae6eed62 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.out_proj/_s.pt
index ddad337051efde43dab778259c59c82b9f6b2aa7..04b67e0ec59ff7a8866de7018b8c4bf342742ba3 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.q_proj/_0.pt
index 02c6a65923229d4f3b54583ea5017915ec712633..2f6bfeb9067b6add2cb196fc520eca6591b16b12 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.q_proj/_s.pt
index e205dce3ef47ddf588ee3499b28433c9e5f08653..973484c792225dbf272c1627bb0684e1388ae957 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.v_proj/_0.pt
index c12efe961b4bf8c7b64b9c7b190ccdd6308d3619..4e13b3b6298d6112134b46b58b88d3b863ccabd9 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.v_proj/_s.pt
index 63be4f775d02371cc952e94e639a5f108f28be5c..808175e4d0a9050438f6ebfbb4ba08f5a37e0742 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.18.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.fc1/_0.pt
index fb83d3612cd4e85a370bcddfb96f701f80803452..c44a5b2ca8dfc0a0fd263333fa7f521b92bc3d98 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.fc1/_s.pt
index 17046e71198d0e7c8318617c537953532e85aaa9..5f9bc1301100edd410ee64875a76ddb785d3bf24 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.fc2/_0.pt
index 2391959a04b91182ff950967318f962131fff6c2..4b67f90223975fd56703593433abd4787876b84d 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.fc2/_s.pt
index fe86f677390283891ea6a8b69fa23acbb6ee004d..d8dab87e057ce35dac878a40167acfa778902fcf 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.k_proj/_0.pt
index ec7d68464a752e47d44067f97d032bc98c4f6a0f..3ac3dc41cc92e4f39b741a61a732a1e09b60b4c4 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.k_proj/_s.pt
index fa01b48e0cbab34b7a59bc32b6decef9b625e89c..9a8e7d36289e59eee5607c06248fe892086187dc 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.out_proj/_0.pt
index 945eab5ae5948fde4cac98d2e0f9ab282c45b81c..f9d36bf5974b68ffc2838ba3c7f571944fbe15a4 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.out_proj/_s.pt
index c01c724dbb01df3e82476a796339e6897ed24e3f..d01623472866c3f42c8e9390c6d2689d6b237b8f 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.q_proj/_0.pt
index 0af592d99cfb047667d7f59f3b7741dd281574d4..28276fdbe7802e1581ae53d565aea3f9282b2497 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.q_proj/_s.pt
index 6358c232467f53731f191466cbde493fe6101cf1..5aaa8a5e8a83e64f821b165ca6a9c3ba5fc22a76 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.v_proj/_0.pt
index 2092bd7f58de3346813de1cea9fb75a04e9fc6ba..13edc7f0ec1be349b1e5f365bc09a91047421dcb 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.v_proj/_s.pt
index 0eab44565767d8bd644a3cc59a40e20bb9f10caf..17db0a5389d3fb1d09a52d0080bb86617a746550 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.19.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.fc1/_0.pt
index cc079429fb2e3c8ab0120767afb37d1845164537..ce1dd99a988691f8050236204a2b025fd64b5a89 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.fc1/_s.pt
index 7aadd91729d33c9c2ae6bfd761ae8e7d0974b2ea..779c43af803da73c5bcc61ef0de3c8c432bdfcc2 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.fc2/_0.pt
index 4aa0702a15f2d4e08869f1722152638ecdf18b8a..f4e70a57a47305fdfe074c45443c626f1569780d 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.fc2/_s.pt
index 6193e3bf9ae18d9b15cde0e660d21ba74dae5732..b66c0efaf731f4d6105a2d1642ae91d11280dedd 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.k_proj/_0.pt
index cd11c94de424cc7ebdcf692dd3974c7cd3016535..cc9c8e44437aa6321d781b2de6c3d49562faa7cc 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.k_proj/_s.pt
index 9c0f27cd44b2974cafa448a86a98bb3070bed980..2b8bae5fddf02fd973d1802f6a43d9232b2ae28f 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.out_proj/_0.pt
index ca44e5c699405af0dac9015c517df9e93c2ec454..194b79235ec075c149a6516a6333fe2c8a7af6d7 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.out_proj/_s.pt
index fe33cde007220c5f5cd905fe333d2982384ac0fb..e6119e3f1131d576b521981f3b125015530fabd3 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.q_proj/_0.pt
index 7b38b69debb31acabb4ce0a8f4e10d62caaa3a0a..dadf8f083ba5e031db018e7e0c06116d722d6896 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.q_proj/_s.pt
index 1c1716825155c1c6b1af1521666bbdbe60fd6c5d..fa989104c788e6fdebab07ed3fb0846a90c39547 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.v_proj/_0.pt
index 24d76dcb711696ebd43d3554a3648b46ee9a64ca..886a9db42939a6ea0517c1cda47186baf8825511 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.v_proj/_s.pt
index 158db523dba634a30149d1306a6c4d531f0f41ee..796bfb31bb55407275ee2b1007353d9c6de3c6e3 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.2.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.fc1/_0.pt
index aeaafaf5bc3c4135bcf26470f554b22a463f79eb..4bdc9227888e63d5dca179b7b5cb0df4f48e7812 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.fc1/_s.pt
index 8b9e78e8180edaf809f058c498b3690dfbc4185b..cac068a08ab03e7ec2724a87662dae6ee8bc4a0a 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.fc2/_0.pt
index ecab1116982505f1d14ef31b265a4e53a543e1de..45bfb1c973e38adab5834eeb3924d68039b4c2eb 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.fc2/_s.pt
index fbc733c0e9b2bea15305c6c69f35b9baa17b1718..5650dca3bf07962bc4134501521540cb3116d81a 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.k_proj/_0.pt
index aba1739be99d79dc5eb57e81143ae3419c080d15..576837163cfb9682f979812897227b5f51d128fb 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.k_proj/_s.pt
index ec5568cbe553e9d3a39790f421b55d912eda8d21..c2c858836be65927ac7df777aeb30c8d687deb42 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.out_proj/_0.pt
index 480161927c4c9d8a4033bc9591a3f53161e0ee4e..98118035ec3a46841ffedf87812bb2c303638cdd 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.out_proj/_s.pt
index c51c4713fe6a73a40564a95a7379b13871d27be2..1c74e322e581931fc5e482aa94c82a0af00353d7 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.q_proj/_0.pt
index 8b35708f58c3e032ab27ff614f2544d67096d86f..51238a2bf0d8c0a812af70a5ea6d160a3d1e1878 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.q_proj/_s.pt
index 60004703636c119a2f7efc25ad788fbf3d9c85f2..5d1ca83d414ac28b5ecfaae815d58435330b17f2 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.v_proj/_0.pt
index 95af6e65ee80092e5143ba78e466cfb1bbc47290..f3a6039172c32dba7f373afff606e55a01d4c39a 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.v_proj/_s.pt
index 906341b3653e84c7638d9d67cb0be14583f9fd67..66a859167790f3cccc7f5c485ac78c0d771e039b 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.20.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.fc1/_0.pt
index c0b2c279146438d9274d239b34102964c08ab2d4..208a1aafce5b2c69ae5e4bf296c686f91a13d175 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.fc1/_s.pt
index 6bec478808841a24e88450911ab28220403f5f86..933a1471edc449a36108762d4fb4ad6acd7245f6 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.fc2/_0.pt
index 7263fd2ec5d90e246a16fbb90a4bbb76e830ddf4..4457cc95022f0946ebb03714ed4773330f4fb387 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.fc2/_s.pt
index 6493023ab2789f0f86d30364250b9dc4b6b121ab..6fda56c4125fc3406e71d6f2b34e59abf0eded76 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.k_proj/_0.pt
index 5435007b6c69e4fa3568e9ec66cf8ed9650e0506..4c118bbc24835dbea892df3d89d715a36153dd68 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.k_proj/_s.pt
index f670b4663a01da325e0dd384dc9216928faa8622..1d7dc517976271ebdc1b14ca67b41ced93286ecf 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.out_proj/_0.pt
index e5b91476f0e344228873025c1a5b41a47bbac887..6e430150b33f6277200da0f472974b1f45a9f7c5 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.out_proj/_s.pt
index 9aa9f513076363bbf8a6c056f7b3b0f905b313c1..b86d9feb8571f796c9236b2132396f2329c487df 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.q_proj/_0.pt
index 1fe87b7721a249401dd89f604cc7dc69748a188b..6771d09406fab03728795f86b5149a47540c3b05 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.q_proj/_s.pt
index 90c2e7d0aedadb0bd5c6c3d7dcbf1d09021e0d76..e7f6ccef4913a301a4ba79b23b29a0aba1b46df4 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.v_proj/_0.pt
index b7834736ebb0b95f56543d081e3934e847de5006..10715a5e0a96772ff8e243ff47673574a1c7877d 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.v_proj/_s.pt
index b5e90858b690acef82c73350b43962e222dbf23f..4bfb33763a7cba6ec999c2c16c8e0a0e50545bbd 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.21.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.fc1/_0.pt
index c15333d7dbb2c79ee5ff4d35377160326b1fda55..3966eb71449fe32b62961ea686239a3048a43e9b 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.fc1/_s.pt
index fe87280e0b82e72c155c89d747bd1f2b994c92f7..4e435802d05767cf21d7fa42b1c3677eb0eb9a03 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.fc2/_0.pt
index bf843c2aec78f00f7579ba6c8ed75ad5e8154dad..5b9994ced953835b3d259f530740e0e2e7c6d477 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.fc2/_s.pt
index 74a23f5dcd31177fc092eff14db9c0d4db3ba13c..a98f25d51e1c13a12644db2a8d343554aa771d25 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.k_proj/_0.pt
index 79bcfbe33daeb6cb2cdf6b4f7e9adebe9b179fb4..d8fc0b03e065f517521b1754ef511cd2e118749f 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.k_proj/_s.pt
index 53538041921baad63a5c90bf19182a0416405c31..1a62fd54976ce9499b54565fa823bfbe30652799 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.out_proj/_0.pt
index 7785edabf78fc691c328870ca2a5235594cd1e41..bfdc6c5ea25a3cb9b9e1a814305da34977c87d41 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.out_proj/_s.pt
index ed6738c1f44d5bde79204cbeacb7e4c8944139e5..738d57d5f40309cb3ba4e08239e57a9104e7a3e2 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.q_proj/_0.pt
index d93001d5a9099fb7abd9f1e156c897493f1cc78c..d3c009a02fbdf6ff41123c578933ca72b6b1ba5d 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.q_proj/_s.pt
index 8ee5b56c94a4b411b974de96b34af2c8078a479e..5fac07d9c6c775331262d3879934768af3112125 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.v_proj/_0.pt
index 0789e4085785c87fa3aad53914ecf139b0898736..ac7ea995ccf648fa04bb962095761ef7970de31a 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.v_proj/_s.pt
index 353f01fe00ec5ffbb41f22ab99edc81c46b8ce54..4d75faed6b9610262389b3fa75f2d1f2fcee28da 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.22.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.fc1/_0.pt
index 3a38f6d40608a9a5dd35a005083e19b6dc84e445..e16de06cabf160c9fab478879414517c7d470233 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.fc1/_s.pt
index bedde5d90b67bc081373bfb2c25628ca56361c6d..3d408bc1113e858cd1be3dec29deae078e3b4da7 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.fc2/_0.pt
index 1719c386ebe598a95f693da7387ab2b710edb372..de2ca3262fd3028530a5b21d8b39a7fbe3cfc354 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.fc2/_s.pt
index e43c39964571785f21b9aa41e2b26746bdbf959b..3dc85faf8c1e506a7c8fbe49e2c1b6d4f1a40f64 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.k_proj/_0.pt
index b681a3b2c7336b7fa8a9768884cbfa74f1983ddd..b0fdc39f0c7950757e7e5168cf6b9502d1ff0a13 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.k_proj/_s.pt
index e71d1db1744d9a3e4c825b37640101049dcd7ece..88797cd6ae7008b6fa60c0f0a5698ab67bf5fdb7 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.out_proj/_0.pt
index 88a762a16a54ffa58d4d9d49cdd9273c4986ff70..09ea0d173a61edaa7dbe5f675a4b561efaa7cc13 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.out_proj/_s.pt
index feddd5c5e660ca4b1c9204b8af539bca9f0be978..61cd483560ed0f9bdcb7bf1403d209ee2ca7f497 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.q_proj/_0.pt
index 927b3f5fa19ff0aca6765c38117972006f3fabee..eceb5a7aa032ef928c47f8e25a3d2aeaed2a7390 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.q_proj/_s.pt
index df91d99d2d43fbf8142ef02d89dfb052a0ca7282..8da5a733e8e096f0dd81991e76a87ccbe9b04369 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.v_proj/_0.pt
index 8b99d7f0e6f7146dd0ac95ae77ea7d5204451c31..954912afc3a0362ee8356870cdb6883c0a4bfbf3 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.v_proj/_s.pt
index fdf3f0c288245160ca9761c3565f2dbacd3ba43d..1fda1cca01de8c64230b8ef545383af4501767a8 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.23.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.fc1/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.fc1/_0.pt
index 476861cd7650baca0782d5e38f1d3099c82699a2..fd2a1288fc4f544426c2ad5e6170b943166b373c 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.fc1/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.fc1/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.fc1/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.fc1/_s.pt
index 9ad19170bc4831085bcbd1782399284b394786a1..e492af561a33d5dc263b3a57dc87b66f08fca49b 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.fc1/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.fc1/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.fc2/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.fc2/_0.pt
index 18c1f4975d347ddacfbaa674d83f5d26f524d704..b0fc10ca98e2dfe740bf2824d676d4b0814c2192 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.fc2/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.fc2/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.fc2/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.fc2/_s.pt
index 6c15a9896c8a2652fdd0f3f128cb2534bda63535..2884ceda279993f1ea9c09969fe2160a1da19844 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.fc2/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.fc2/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.k_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.k_proj/_0.pt
index a3fd5bedc125eb03a19ba4f00b3b1b521b93f094..687f0f3f76cb70b8e44e0be80336ef0559de97f0 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.k_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.k_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.k_proj/_s.pt
index a810839f1d3bfa677edb0ecbc75098de0fceaf83..a49e826dabc3ccfedd32282571aeee9f9bca062a 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.k_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.out_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.out_proj/_0.pt
index a017f6d36fc5514cb7a6146d65d3e38d1832e048..bd0ecb7314516929db5f293292d33221846b42f7 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.out_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.out_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.out_proj/_s.pt
index 6084244405687f4d27f2e9961d707f659f6f57a5..0051ee7fa72c0457d8c68561581bff0d52866eec 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.out_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.q_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.q_proj/_0.pt
index 02919bacc399d9deb6770dd1eade0d5ec632e0f8..dc33ab5499f4b8aad500861e4423854bf8ca59aa 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.q_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.q_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.q_proj/_s.pt
index 4a57408ca15295d75b4b3ef5ce8f4bc09a72def1..84491fdf8e80e5ac4858b64d76f08543cde075b8 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.q_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.v_proj/_0.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.v_proj/_0.pt
index c636690f2cabb4cce2336a3d4a4d2d6077119d4e..794a26f6e902ac3d48fe13e6db2eabf791da8b33 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.v_proj/_0.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.v_proj/_s.pt b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.v_proj/_s.pt
index b88880d18b234bf0e599e4e101d9eaa907e6765d..76cee4e5487ee6bcc7eba6ddcf8f259fad65d853 100644
Binary files a/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.v_proj/_s.pt and b/params/opt/6/rnd_search_t_cos/comb_12/trained/model.decoder.layers.3.self_attn.v_proj/_s.pt differ