diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/lm_head/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/lm_head/_0.pt
index b26634d31627d1a0044c3f3651ff144ed5eb8f4c..257f78e580e29f2b211849ea7bc8ce9dac122cc9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/lm_head/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/lm_head/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/lm_head/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/lm_head/_s.pt
index 5f7cd275bfa5954cad17f78eda73e6fa68a1ad13..264bca846916d0a64a33029e9e83100b3a4216c5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/lm_head/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/lm_head/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.fc1/_s.pt
index 5b39d0a4869b36abe82d2dcfbe9dff4dba3450b1..a695dad95f8facecb922700fc90a0fa8a03b83c3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.fc2/_s.pt
index b9911d4066649176e3919cf3807a456125965744..57e169cbb889a1d20c4abd3681e01648f949fa1b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.k_proj/_0.pt
index 7c8c0468a250ee66129e2cb2579ea41b30e11b8d..88dd9192631cedb10ed1a0005e9e74cc4b5fe600 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.k_proj/_s.pt
index 3614baaa54c944850c9000b294331edcbd364779..e5053fa87e3e2169a8792130e04b796efe89831d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.out_proj/_s.pt
index 35a680b769d455db102232ea2dd8a950a91d010e..dee7624a8a865219ed8274eef080414ab70b8b2d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.q_proj/_0.pt
index d29237a9ba4ce901679ae5b7cab68f9ea0db14c0..1d1ccf6a613d81d3a137c38a420ced9052d940cb 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.q_proj/_s.pt
index 8c753f866a2c728631afcce977dea9eeabd82bdf..a7563d67bbe57364f25f2bc834863ca56afaeec5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.v_proj/_s.pt
index aa5844a664bb85fff376b5b832aee50d3ff6afbd..db0a3cca506ec7a5d7547411ad3e849eaa2ea39b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.0.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.fc1/_s.pt
index 631d37b3d78ae4706cc9381814e8cbac2d9fa2a9..e8e3d0168d6b965ab4266f2c89b707fd565e5b19 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.fc2/_s.pt
index 36a4d98e07a375e969b1edf2bc15034d681fad06..65b76fb7b01394912998c4d984e6424bfb740bf6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.k_proj/_s.pt
index 09eea2d74cfc13fee66eb35652d85f8520244303..e36768b40f990729ac8b28115867b886f632f37e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.out_proj/_s.pt
index e11e9a98cd8f65338bfe4f9dcab355ea98d7dd27..1b07c02e04fbfc2d59fa4296b26d47676fe89db9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.q_proj/_s.pt
index a2d6ea5166fcd9f63e1a05d426811d38429f7117..8786775099a08f94aa79c3d0d5d7563d01644dfa 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.v_proj/_s.pt
index f21781afe9f2839ffbea2eb0039e324e47e065f0..828e814149abc249f73132aedf4b260592fc3ad1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.1.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.fc1/_s.pt
index 513f311d038e36ab0f4a2d6b7159a95d5eeae9db..8d6cb967271e02032cb0d6f8d3026612001c03a5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.fc2/_s.pt
index 3a28a13ae9c865800e1f5490a6f18f6de4f370f2..bb54b619c8b4ed57759a38a96cad9238e7c991a4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.k_proj/_s.pt
index 62b269d4dc0f6da26f68b023642dd2a7e21e68aa..cb850bba533aa39599d4facca65765e2ff580597 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.out_proj/_s.pt
index 077b92e2b09d89d5793610a8861124e162ebfa6a..a4d35426ee2861428fc630e80a5e34c33ebd5b12 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.q_proj/_s.pt
index e2724e80fb68a42dcb43f323dd59bffae3ba074b..c993a99888acd559da021609e26d24c5a6ca3eee 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.v_proj/_s.pt
index 98a539b1f5c3d33b6a70c63bb332eb6ab35a96a6..2d7fd14e3f534fbe21aca65c166005def75b65c7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.10.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.fc1/_s.pt
index 36c474e7d8b00b3ee0f2502616683ac52c7d1619..8863b9e21ffe43775e90d216250ffe30e425aa7e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.fc2/_s.pt
index cf7e36eb46a34be093d03b77d8b0c01bc92d5930..63d2c4380dd19ae853ed369826ee8a003921e656 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.k_proj/_s.pt
index 5d2b2d82b6c5a842bb456d1f962284928a5fed44..f79d9c72afdf35e892f1288fa4488cf30c127c8e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.out_proj/_s.pt
index 5a028df76e51925800a0f9fb7a78231852ede5c1..1d245949e6150147b8ee005b8bc7eb62fcd77133 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.q_proj/_s.pt
index d92f5a5f33620334b89c6b1c247cb468293c5ae0..c1d61b3382aa4bcfc30fed276dd6e00eaa744dbb 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.v_proj/_s.pt
index 60336255de519728427eade6dfc71e189ab150f3..bbde6ee679bfe765d3d949696e4e8b21c780a232 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.11.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.fc1/_s.pt
index a3af70f1ce3717767ed8cab1101091f54c7343af..36b0a3d4fa8efd13fdbde101b28413bb3d4358de 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.fc2/_s.pt
index 1e6890da4898fe37063bc8f2a25efff5f6da6d6c..50bbf8dd853e9c0e25675e555b1ffb5e095b5670 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.k_proj/_s.pt
index 808a9ea4f998acad56c5aeb0264c128aad7bbd8d..7b946a177b2aad1eb45288e440012378ce0a533b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.out_proj/_s.pt
index f17b769488e17acc4abc100a6776e6fb2726f0a9..9e3476d9fd68943e4be6d1f83ca4c805a8fa82e1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.q_proj/_s.pt
index 6345187caeadaa198c26da44ab899218f6476853..ef8657599610a630dd38299f0d1e7c06fc02c981 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.v_proj/_s.pt
index f6c26d4b725378acb1642c3b8479e887b5d220ff..3ba1f76b87797c686526d140d71c44e4ab8b4891 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.12.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.fc1/_s.pt
index 29ed76a4c0198d7513232cfe0da778927332d1b6..47c3443b7b096d8a83c8244924913d34735f6728 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.fc2/_s.pt
index f34087557a7791a8881358f81c6d954dd5a68bd4..6a0dd8dba47ded530b0d0fcf94d40e740505b1c9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.k_proj/_s.pt
index aa4ae03c0cba7238948e26ea85cf9632fe8d677e..4e9f38742142e9e7846d6ba8f29e6641b13bf043 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.out_proj/_s.pt
index fbceed20e91f55f5060772acaf1884a638d54e7c..b4614010422b393f96368b2293293c7846711f40 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.q_proj/_s.pt
index 109111bb79c6c1c19cb49056d51f345b2a79fbf8..1c02933d7e5b5d3d95c4656df285be164d464ad8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.v_proj/_s.pt
index 24bf796406378579885c79e8e7f0c76b5c338dfb..030f959464b3992220c769ad8176527a435ffc5d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.13.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.fc1/_s.pt
index 131c3da4fa3ddd37193145c3ce8c84814c518404..098c96f449f39fdff44d826fb3226efb0d125caa 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.fc2/_s.pt
index 5e96b3118b86097cbc0497bdc4581b20a6876f76..849c22374281b2e57af648ac33450f17e9bdad05 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.k_proj/_s.pt
index 78718bc33f35c427238df586056355d5a50be1d5..34b7b5fc43ebca8b83b5bfff463d4f98cfc48528 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.out_proj/_s.pt
index 20f4d0b6f6644df7a7164874e63bb743a61ad039..851a722d8d6737b4bd6fa141ecda545da420e260 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.q_proj/_s.pt
index 1fc35f6a555d0950fbaefbf1715d939b7ea5ccf4..1b803fbc1997912d07159591c3ed790e93cd20b2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.v_proj/_s.pt
index 3be466da849952c0dde00776dd1ecda25548b819..59a1ded8750862a879792d6bb87aa7b1b79cad4f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.14.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.fc1/_s.pt
index 0f30a5ffb1bca2466f2524cd0f6ca83708c0c8b2..1326ab0415bd21e2e039af3b0b2876de46d393d3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.fc2/_s.pt
index cd734e5f4a30fc7542408bd61963bce0816ad043..fe423278b62904e34833e5289081afb774813cc3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.k_proj/_s.pt
index 508290b774fbc6ccf96a3b13af3b38df0ee32678..cf92c7083f5b2dd7697bff2680c0cb16a0aad36e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.out_proj/_s.pt
index 41ce5465941e7ff25320ccbbebb929ac45992555..92d7736d1e65bfea5909cebdb4d49ecb315d464f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.q_proj/_s.pt
index a6ab34cfbfb0bed1b9a7f07753801a629e984031..0fea541b53121abbfc93d1614fd8b7d46471140f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.v_proj/_s.pt
index 2caa50af35826a291b7eea2fc01331f5b9ee289b..220bbd2dfd088a4e1b95fe3a3e720ce2f2d4fa63 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.15.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.fc1/_s.pt
index f5a4753c81944770b4270d80709ce3e8213cf549..422330478341ab1bff5d962ac2f44820f29cf397 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.fc2/_s.pt
index 7df837dbe0377f15738605ab0c555f4981af5f7d..2f7b378a531639db84ce05baef017b6c378c75af 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.k_proj/_s.pt
index b31c18b6328fde3af1c8f81febfe44ec56a7b135..9f164fde57e3f56133667c8828627a96ab1dbc23 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.out_proj/_s.pt
index eead92d00b2b253980ff4f5a0f62afedf386d55c..09360b0985275496fc61c1b5bbec01bc0ce9c8c5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.q_proj/_s.pt
index aff59f8cc77a78b16eeb452d03fb5f539b683d86..00afffc59896bbb54211be41a4880cb56fca12e9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.v_proj/_s.pt
index 4fadf688bf88317ac49869f5b1e9b81a29f52916..fe7b02768f1b3cf616e8804b204137b066357169 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.16.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.fc1/_s.pt
index 7aba9f90340a93d03c1836e35d3717e5489a3e63..3f50fa1e44295719160040040e4e01200b8e82d2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.fc2/_s.pt
index 850657257181cb84a9e282c788a23ad7ef4cde5f..c7e376a3c8c8716c36284235dda8e32ed5087600 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.k_proj/_s.pt
index 743d9de2c75b5d5bfd1bb31054bb5c8a3ee6433e..2dc7fddae2d105ab7ccd7f360ae19fdd10fcf75f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.out_proj/_s.pt
index 3f15ed04be81d3c364f05e158b93560d7d36f899..2114eb6e96529e4faab794cb472d48aa915fbbcc 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.q_proj/_s.pt
index 3b436ced7ea594250007453a27351c8e45f69b87..b15a46fc02c67c9defc219e50cac65d35c83ef4e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.v_proj/_s.pt
index bfb937a7826ede4e4e9064ab0230c6bf9befb59e..ffdf6c45d76160776a8c00e9c6c7427f492c433d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.17.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.fc1/_s.pt
index 27611f5242a3d47e4c7e6be602a14deba8cedec0..4d42252dbb9461a0a2c67a1fc736e775c15e8e4d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.fc2/_s.pt
index 6a4388c9e91fa965a680ed60495757ded26a9089..b85fba64e6d984f170bdc8dabef7bbbefed3b3ea 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.k_proj/_s.pt
index a5ab88528b67927cb151f89e1c7278f13c8f1950..f7d2d97bef41b48f3e1092d2f4cf3a4a2f7db582 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.out_proj/_s.pt
index 665f281589fe6fba89447d7d9ab833fee0234b5d..ffec1a3fd6dc0e4ade984c89f59471fcefbadfd8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.q_proj/_s.pt
index 2be6bf7802e8e0cfc01c6431d7496628c1d633c5..095e779c195d99497356535ccfc5efba9f0f9a5b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.v_proj/_s.pt
index 78b1393c3d701b29e8509b6181887ca828150a7e..d54cd2cfabf018b5bc6e2f65a7dfc5e4cd07d3b0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.18.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.fc1/_s.pt
index aef1a2a10e90717cf4e0ed7b85dead16fdbc199a..5ab35f47c47d27a91de38309f7a97ccbdb18190c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.fc2/_s.pt
index 242dcdced424939fce37bbae9e80c0b978a17e3f..4997a6f9cd5d02971db54787d9462b8d8569151d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.k_proj/_s.pt
index 5c5f5857f5abfefe22f1e3ed9e58154122411c8c..1531e963d9d801ecb5e70dd738043f69a2038bbc 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.out_proj/_s.pt
index 430c6d7c02cb4cf57af2d71c89ccba1fef342f6e..3ec0c63a8a5ec6cfcc303da525db564731e08125 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.q_proj/_s.pt
index b26920c4a1de6129d8b6c4b0a4bd911efb20623f..5cfcb58d33b2a4cf818f525be350efb59703e4d8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.v_proj/_s.pt
index 9bb09c50104cfa346774ba79dced9789d6a1dfea..e70786a81ac6168909169ebf554d5887ef5eed9e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.19.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.fc1/_s.pt
index 5951daa1d07667f99e76a7d5470201903b59d0dd..c731183f9beb84cfe50b601a4381d5602764891c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.fc2/_s.pt
index 1ac2352c2b5a40d475498c59151fcdf1635c4b37..c9454f479ff4ae23dbf9d27633c36efa6f9504e7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.k_proj/_s.pt
index c304a3dd7888ae36cdf1a1ffd95cbcb961f9b304..ae3cf9f66684cee0c9b9d24f1136ff1739f27f14 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.out_proj/_s.pt
index 41da7f27c47ef7c3beb1c1b8da5b8fbe5f6a4901..fbd98e2a3cd5d790c516398ca597dcd9335714d2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.q_proj/_s.pt
index d627dbe5616d04846912fb8609e5bd376f4e5b44..58ce624a99f482d520ecbb9740eed763909f8db1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.v_proj/_s.pt
index 967d16bb173136d03240e9fb250b887bdc5eeba0..0e38f553d88936b2678d00496816faf662dd7e8d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.2.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.fc1/_s.pt
index 1058bab34503e1ccd0bb3e2360ee611d106ce92d..16c9b065af8cce8b4ff4f59218e3419fc345b38a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.fc2/_s.pt
index 3b5f68e1257d1b9da11a12c1a52ef9e846e1594c..6691e3e8a1dd1d8ec689a40420edacc597617b80 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.k_proj/_s.pt
index 476b3e40229f6727e9855e630410665d3753639c..8fd9ea375645aad32e49330c9cae23146f74dfd6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.out_proj/_s.pt
index c803143f8a4afcfebd90b99a9a9b4c7a609cdab6..3c1f9bb107fe2ab883ebec408735e37cee4738c8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.q_proj/_s.pt
index 8eb0ac3d479c024125faf118f1669dec4be709ac..47221921917c1971768fdd26a189234bd0067f2f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.v_proj/_s.pt
index eb84c2cede4f362d53dc907957e82dd7d31c4ea4..c0d750605cb1e9f0798a50041cc3d35ce1e71d75 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.20.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.fc1/_s.pt
index 0e20c5af4dfda022e6e46f096b1b2fb00244079d..a6faa0ffe7d4f6ca5fc97516ca6c28ef24123a53 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.fc2/_s.pt
index 8e702f09fc3f5fdc175531134ce9bbd93944d6f3..f67ac15daa6e718362e53e5820ce2d3eb6ce44fd 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.k_proj/_s.pt
index 352af342d2b3b7f8ebee3fbb5b22284e2c4eaef3..bad1b9a640b062dd35c6547d931d6a065669c345 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.out_proj/_s.pt
index a02040c26f91e5b52990594dd9013e7aa3f195b0..8d5086bdb6176bd2f2dadd87970b30e61a2d2773 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.q_proj/_s.pt
index 2b4c5f243c0be5257dc2b01a76679c255c10973b..8fe35787339e0ad71c3451a0b528d0cee412505f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.v_proj/_s.pt
index d525cb33e798ff9e3e924892e6d16b795d414307..7f51464d2d334af18e1a3e1f3fa4099e1fdc57aa 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.21.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.fc1/_s.pt
index f0ddecadb8a9b866ff9d68734f98a48482f51d8d..a1dcab222aa7addf055053ad2bec0fabf46a1e01 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.fc2/_s.pt
index 5f0022c4bf0326c921b6ca2b23c02935b09948eb..def664f40d973f20c9eeea00f264ed96c1c24da7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.k_proj/_s.pt
index be1257e00af015234859abf43ed76a0c518bd031..01c556cb6c7272a7b52ed1dd9b92ca2223bcd236 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.out_proj/_s.pt
index 5b0ef61364b2c1453e42c5bf54c64a31e5a68d8f..e0a8c73a8b3f3ebd4cc38c8005d38c11af9934c8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.q_proj/_s.pt
index 6464943f505c39f7d82e35e3a22c51e3d1177b40..34f4f87745cb92c715e5eaef1079f4a3364cc217 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.v_proj/_s.pt
index 1a97445d442dfae9478ec243c5ab7cdf5e4cba18..9bdf83e61d581803b88388b8962ccfdcbda8c8e8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.22.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.fc1/_s.pt
index 5c179ec37d679714fee5ceb43c137e36c3d2996b..389008a5cb6fdfc7e433c1d8285b614af2b7b534 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.fc2/_s.pt
index baafb64e3e1c83e7a31f88b2b8e1b58ca1cd5467..1c844f66537cf474aca45e88390c841a00735aec 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.k_proj/_s.pt
index ccf3e43d0fb551ffbe452234b93c37703c040544..40fd72ab842351510041e9118ff17b99af8be0a2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.out_proj/_s.pt
index 7eac77198f864cd4c9e5a1fef55dce18692ddb1c..a8bcc6099c18e8068597172a5f9b4c33b12060ea 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.q_proj/_s.pt
index 5dc5628210751029ad95017199b4a501f56a6ce6..8e6ba3515700801dc0b680d13c8855295f7f5f51 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.v_proj/_s.pt
index f9425640cbfb405da375374cb2c86c6c8cc1da49..1c2ece927c58003e6ef462873ad18ba8dede1cd3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.23.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.fc1/_s.pt
index ec1f0d55ee9a28d9b0c89ce2ce0693b0a8262a4c..2b0a0c3e1224a4569bb890a927f04a83fdd22ea4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.fc2/_s.pt
index 4680b2e3d36cfeb927a4386013c8f65969e024bd..9de933c50ec06524c89b4c09ea60a39b298a4d79 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.k_proj/_s.pt
index 22d6ccf4eafea5c29f181a239d6e17137512bbd7..5f43c919461c212facb0c0c573fc3509d5c791cd 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.out_proj/_s.pt
index 04bc2daedb66724b735947c027fd7e9b63b8ccc8..1990885df98c0adeba1c67500c97224ecdfe0029 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.q_proj/_s.pt
index 79146b96e079875fa9498d4f9c639bb89970343e..481ce51132dcb60230213008da033bb70799c832 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.v_proj/_s.pt
index 2ed3b62170ac8c68e9ca4eda51eb3710eb3e7d6f..717a902eafded6534240e5b780fcb7455e3dcac6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.3.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.fc1/_s.pt
index 8c0d35bb70044bc90b490d80fab1bcec7c67e2ac..3e5c23ec83b19aa7b793dfb42b7fc033e3733e78 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.fc2/_s.pt
index 190bdc7cb8d55adb6ac6babcefcb5e6455fe3e0c..a2635b64a316c922afcaebba4bfae16a8a0b397d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.k_proj/_s.pt
index 4c96fa15ba1deeb204a4258cd3392dc70a06a799..8c614115eb2260658a42f95cc9291a8deb812b19 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.out_proj/_s.pt
index 95a7d7cd6a1565cd04e9f5cd52df460a85143130..5d524ab38ecbcf9e949236652a34b45daf54114d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.q_proj/_s.pt
index 091ca803342646e96dcb34d7c4a4bd32f854a46f..3d82bb21e4dbdea0ac14de8045852eaf84ed5c7c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.v_proj/_s.pt
index 1a6df6f5ba421ba77b15caf427499d6591ec9ff4..8d2f789e25dc1b28f4bf47f1b25bc77f734735a1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.4.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.fc1/_s.pt
index d28f55d2d425bfd79423454c5a3bb50a2984f4e3..b7717defc0310fb6de8e68592447cf3010ac9c2e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.fc2/_s.pt
index 06f71fb4429fb30ea4f169d0dec5236b351e32b6..e91391b19298d42c528a6337f4a2200484cfca45 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.k_proj/_s.pt
index 8126ec49a4c224b0ca5ccc4955bed2d89fcf7dfe..fb9286a6118b2c70e0d013dedcf1cbff792f1065 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.out_proj/_s.pt
index a9ca8d836d2a7bcce2043b620c43e6a12696dabd..8586342c2d2c0c150f4dc410f4033b83d1055016 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.q_proj/_s.pt
index 6171a5b51bb8c0846338f1d15a545e9ff6722e6a..bd5e60bdf18fd7564bf1f4c97ba999cedbd3f7f9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.v_proj/_s.pt
index 1ef47c41cb70055b83b950400d742bdd93cdce1c..e426a4e3267e25dd92bcb69b9a37df687f9c4342 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.5.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.fc1/_s.pt
index 4cafe20c01e8729f1c0cb3cf7a50089d619383c2..365a8c26a8587811f56c2979714251b313ac4b98 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.fc2/_s.pt
index 0019e1ce3f4e672dd8712c16d9b51971fea4cfc2..207bc9278d8cf20a0ccc2e68c4be5ba6acf2a685 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.k_proj/_s.pt
index 28f99ba99553fe1c54df84d57ac1672a87b3e82e..7bf8d048a360e3aebbb588919f29bdf532c90508 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.out_proj/_s.pt
index 28532132b6b91823ee77b8e9a892f823ff25f1d5..bffc62533520ae92f2cf7d0d5039fc782d1c01c9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.q_proj/_s.pt
index daed49ad1d8baefbc1d4d3e1d3389de4abb6597b..f5109c9042cf57e9d1a35ee1413bdd1114d668e7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.v_proj/_s.pt
index ce52bc447d888583ccb7e83abc37b5fb1eede54b..23f2048aa09d1e01ed6dabe80588ef37547791b7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.6.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.fc1/_s.pt
index 3cbca6724deeacbdfe866d63a75917b28c92dcd8..63248dce2f014bdaa23f0bb1d0643396ccd3fb4b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.fc2/_s.pt
index 39d339a726e630811e0946c7620554c76936ffbe..43bf70de64721d916a4b7b80b357aa80dc1c3472 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.k_proj/_s.pt
index 993e5f05bb1085c234e37deb1d3e12c22bb497c3..10ac7b83b5f7b83cc9f8afff803c628bb5d4433f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.out_proj/_s.pt
index b3f1becdee1cce32316a87ce8de17fb81931920e..274f52dbc02ad9c21229ce9682a48bf4f9c96b1a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.q_proj/_s.pt
index ed2073dce980b510df7a19046861078da34d475f..bf46471a94593477c98877b67a07b9dd1edb8588 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.v_proj/_s.pt
index f878e8ad32ad46c3c25f2804c59eb55cd014513f..7552f888a49a375ea29a2f5bbb5b61a6616ee25a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.7.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.fc1/_s.pt
index 72da0a44aa5a55f5dc4a2b792a42bb0fde4feefc..70255fb666bb57642ebde3d70b916701a1979146 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.fc2/_s.pt
index e25496ec18d84b7b59dd3592ef142e9f05fa98d3..ffb2cc1b91b93fe82c17ffb949dd470c659c443e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.k_proj/_s.pt
index 2d3cb674db35053cd999b3aae3c4140c6c663051..a76e471911782d09c03ed9d39c47a4118e9af3b9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.out_proj/_s.pt
index 19f8402c16ea68a1f9cf2b295b150990e88c6d0f..f628dc6cf34e6ac24d2aad857cacffc5e1b6a745 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.q_proj/_s.pt
index 59b5fe9c3c118c2df05c45aef12b05d7418bb61e..69ff00a2790fd6e78b8ad4c64e52382e336d013b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.v_proj/_s.pt
index 9b5778b939ccb6517832dadc3583c1f5773c989e..6fc9a2647c48e6678e4e380238d2ff35c9490a4e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.8.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.fc1/_0.pt
index 58c96248fd00550cc15d645205571ee67c337035..446eea5860a21090a5f7c7578829aa7ea1576bc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.fc1/_s.pt
index 33f282fdf6027f889baccd8c89df76e38dbc0169..0c5845bffe55d6bdd26f192d381ca41a0d4a3526 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.fc2/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.fc2/_s.pt
index bfea38f49fcfd3a09942b94d885f58f6aa7babd3..4db31d738d3a1ec5bab6b47fc478542e5106d46d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.k_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.k_proj/_s.pt
index 71e9c4c117c383e353fc5ec84600838b625f10f4..8a574ec48e2ad99489f7165dccd9a2d9325fac58 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.out_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.out_proj/_s.pt
index 467a398abb9cf6ee12fadd3bd1cf82ffe83196a4..a0f21a17d3fa4a704eac691fc3eb20f78bf90c78 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.q_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.q_proj/_s.pt
index ba702e92b3875c7c9b39f095c121b1a1c9553041..5e17381eb3a344dfdd7c58ad7427c3f5efec1a1d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.v_proj/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.v_proj/_s.pt
index bdf18e1b98464100f6cdbff2701ce31dabc5326e..b740e9536ea2dc22ea2f714b6e68d896ec8f8e62 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.layers.9.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.project_in/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.project_in/_0.pt
index 8ea38b2f931134a804d1dc0f42d8395e117dd20d..7a571786dfe022608d7168a415a64654991d2549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.project_in/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.project_in/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.project_in/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.project_in/_s.pt
index 70acd52a41e2e32e882362b7ed5a6f4f62d35872..209ccbfc507ebe737ec942a358cecb38c4bb2a13 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.project_in/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.project_in/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.project_out/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.project_out/_0.pt
index 321072ecb5d4858b7d9ce1ad463bace6ec840e53..9281ad359b7a8652162b5bde6a613f8ffad67a0b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.project_out/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.project_out/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.project_out/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.project_out/_s.pt
index a601ea19e7a54becd8a2cf4dbea92d05d6327fba..77fdcba094285e59341396a04dcefa828ddc97c7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.project_out/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/init/model.decoder.project_out/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/lm_head/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/lm_head/_0.pt
index c8fbc9da29b9e9a010902194064c41c36f38c226..3447c4ac20625d7bf0051e8f70ef0e68ab935a47 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/lm_head/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/lm_head/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/lm_head/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/lm_head/_s.pt
index ccfee1bfebeecd1c33d58aeb02f598ad62f14d37..393971bf7a8a37c33768f4e34c0041a3b32d9e4b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/lm_head/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/lm_head/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.fc1/_0.pt
index cb97443a96d255cb8377c3a9a893187eb65634ee..9196766cfe576105c5831d493e8e0e74d2aa42a7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.fc1/_s.pt
index 3d480b7779a00941a48922686fc02664ddf2fe38..f2fdf8df18274f959b69721adf5610afd58ee909 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.fc2/_0.pt
index 4c50b956c736872ce55c4e9589fa87cdff945434..88beb74d069f1e64bab0d5d500d9604d5593cc8e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.fc2/_s.pt
index 9396c9a9c932e48e0c05fa2650332331633931ca..bf326c102e1c6af21a8f7a36ae74d17eb7509549 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.k_proj/_0.pt
index 705f7df8981dbbfce40dc4f1d33c775adab554d3..b471b5c9d97f43eac4b89102334922804a4d4c6e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.k_proj/_s.pt
index 1e5c09127e4e791b403c247febafa5e8b4dc111d..8577b376ca0649b89df14e4c64371188b59f8f18 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.out_proj/_0.pt
index 0782ff10bf187e4a7d6ebcaf171fea2aff5b9b85..f2cb5cd68b6bf44cad56c3ebdbe54a986c2e5aad 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.out_proj/_s.pt
index 22ed2e55fd81e502896fab3e0e8e5763d5b2bace..b40641a01f220a554b4c0e13b7fe49632cbc0216 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.q_proj/_0.pt
index 36d47d79231f428ad1eba846bc26562933220749..c80a5c16d957cfdb3794cca5a0f1804a6e63da2a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.q_proj/_s.pt
index 46dd5bc8e80253da8afd3a4b2450d2287afd4668..65981920cab7df6a9753220984f2a1d5ef36e8a6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.v_proj/_0.pt
index 24731b09b2ec418cdb93f20551f0b9ba9fb4ecca..05ef3e2b87de3b54176465926e836bbe1df4d751 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.v_proj/_s.pt
index be2d9274427ffbabdf39b3dbd55a72cb867eabaf..83c612fe425faa1480fb4132964f9d0110db8e29 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.0.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.fc1/_0.pt
index b078833fca26d464ac575f069c04790fb3ad7369..07b6ba63051f0990c65d4596a238a643a30af9d6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.fc1/_s.pt
index e47007b1795ae502728933f9e1056d7bd262de5c..c4856ed4af3681678147d7b407d0612a8aa137f1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.fc2/_0.pt
index 8f0773df1daaf607ee7ee4627ca54fd2dd35f95f..7cc9ffc60939a425efe2df2e637633f7faef5230 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.fc2/_s.pt
index d5f29fccee89b226d5e165647c21d40f0abc6c71..9bedf7b4d1d5bfd6cfb780f6cdd31e56a6d7e981 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.k_proj/_0.pt
index 05cdf3d603696750562448128493b4622a767bfa..ef5a89462e5a969b8d331b950d4b61c52b160fbe 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.k_proj/_s.pt
index aa1ffbe2169bd4213f5db8cbf22d9a24351e5bf7..82027007dba67312d18deb438666edcf1d7f647d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.out_proj/_0.pt
index 45d23b0e6160f997f6c37f5b399ed326ef39a0c8..ec7ae9c0fbfd7a7643157c9eb9c284903172ed7c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.out_proj/_s.pt
index 99bbecb574f45ccea3f137cabc65d959b3af8933..7a64c5bb686a20d95ee03d4f4d91d695b428558f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.q_proj/_0.pt
index d69907f8ef8e8d3bcd93ba798a6d7d10636eb8c9..7f7d104bea2ec3ad37203f803806ddd9a54fd2c6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.q_proj/_s.pt
index 8a71455ebdee30364f6ce4d11a31812eaf535446..27259a199c92ab70189df64288351e920c25a4d5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.v_proj/_0.pt
index 8d725701d208295e493f1142c92a330b9e278daf..845dc6d1d9f96a2328bc3521b3d0d8e3f7e83f10 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.v_proj/_s.pt
index 32444cc9ee99fa245dbc5c342bc6016a42ea9882..90ae4c148e0400cc42e725c2f6d72be09a79ed29 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.1.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.fc1/_0.pt
index 37f8d8cb3deb368b1f69d100e20c3684be71483e..f60c98b8b264eff02d52878185a28c4140c91a34 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.fc1/_s.pt
index 8e08c07aab1d85d4253f3e97215e9816fd0bbbad..4c3f0212ef454c277a7c12aede76a939adaf0a41 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.fc2/_0.pt
index f5b35afdbb05703751843f5441e6d80f0fd38840..64226e2612d248298a6600e75802788aad83a1cb 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.fc2/_s.pt
index 45ff984e326351c517c1bbc6a9f851d95cc489bf..2ba54c3641ef175eb2e78579b0b2f78b127c2b5b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.k_proj/_0.pt
index bd0efb81fe98e86eb28ee95433517fb8eeaad04d..8565b471931d83be5c3f924785b017ace758be07 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.k_proj/_s.pt
index 1252e2b53d36ac691e240c17ab6b2eadb3931df1..455e99d255a135c8ff037ae79d0e8270cd965b9c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.out_proj/_0.pt
index da912d9dfb84d22f684d9aa91bef29b03f6c6a7f..f96e0773e20016f5c96f52cccf92fb3916715020 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.out_proj/_s.pt
index 3c71a31ba4a0f7c03a095639e11c795edc5b25b1..cee426002bae1d39ae10e3cf2bb87a0f5620aed3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.q_proj/_0.pt
index ee1b6e2d261449e86ebf0f588bcbc3b3736ccb6f..2f3e5bc9265f5c3a13d7183cfd3393f09b30d138 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.q_proj/_s.pt
index a3113ff901f7bfd3a7408d69f7d9b06c518fea94..b9ae9fd450cbc9d7bd8b17fd83f8324b98699595 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.v_proj/_0.pt
index dc95f06b1006d38e9bf79d05d162350d6487377b..5ee4601c2113c258798071ca337ca8e1f296dead 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.v_proj/_s.pt
index 22f29f7717f1d30fb5b0ec1c8a9175e1b859d0bb..0c7d10773a4e1f11cc3f03fa1bfe77ee35bc30a3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.10.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.fc1/_0.pt
index 85812dce9b32d0df53c1b10b88960c446ce0d7c2..21b2dd751af8cc93b445bc478ce6315e1f780dff 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.fc1/_s.pt
index 60c52159599782fccb3b6c6ed7a24ce27f018d39..9ee3e9cb0af2861d0bc1ea8cfe888e868e065a8e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.fc2/_0.pt
index 605c49e78c79a5eebaf8784bccc227efac4e13ab..a4be22ce80cfa2df954eaa73d4d8da1548ad351c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.fc2/_s.pt
index 4e05682e12a8a26d50fb74f50c2cf26533f299eb..cc66029e3464e26113e7662d62e84a8064cde3d9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.k_proj/_0.pt
index 5c6ee2248e6b110d8c004e12f3f6a379b727436a..b48b2eb9a9a425a3e9cba118fa41884b26b0927e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.k_proj/_s.pt
index 75287739889694e24798b6f23de8cdb32f019b08..e5c1778d01e937c6288cd9e471e666450af0ce4f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.out_proj/_0.pt
index ca800bdfbdea66b3afa4b15b2449d62cc17a4e40..f024975dad1c663cd440c18d00d295857ba762bd 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.out_proj/_s.pt
index 4f262e8b2313787ee54cf9dd8772073606391e57..1b8ef40e93bcc3090ac6db896f380ee8d9adc239 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.q_proj/_0.pt
index 00c534143338ae90d86de93e0e906549b5927971..43fe0604e4b9ebf9a1ea081565e5cc67274a318a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.q_proj/_s.pt
index 48b30d19ac716ae6fd8ec0e7b0c1ebb31b3f9308..e77501a5bb700548ea0a4fa4b6c99666cc6f0d99 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.v_proj/_0.pt
index f177c4336193914d01f51592ccd5540401a9750e..38a3351880521486ed233542bf2248a2f27d3551 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.v_proj/_s.pt
index dd6ecb6b01a0155998ea89b6916755f6840e0777..8ea88907721cbd8b9e95a93e7d818d11d4a6da1b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.11.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.fc1/_0.pt
index a7e5b16e279235581f17e66f90fa723ac27ceee9..5cbb8f872db3489b456285d1847c515189a93ad2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.fc1/_s.pt
index 6c9781fff293794f0b0f2cbcfa674c9ca17b59b8..3ab2e43edf5e18cc59977be543ae0b09284f9c58 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.fc2/_0.pt
index 5c7ff6a6ff830b17a651298b555b572c22eb18eb..c694d7b5793482c5f34a7a79147c9140a2094ec0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.fc2/_s.pt
index 6cce317c5fdfd9b8d7afa708eae522bb0965ce2f..1dd94dea2f955bbfe61c0f68b47f43c9f8bf4f8b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.k_proj/_0.pt
index 623a913ac71271c95407148fb3961393de171ead..06732f8d919d4ae3f04a43687f42dae23e6ffe2c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.k_proj/_s.pt
index 8763d3a34226f29f669ab8ef963a0043e9c81784..9ec253248d04d465c7298aea79c1ce930a10e22d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.out_proj/_0.pt
index fd5f0fad5e4edda511131abf6e3986cd27d88647..093dff20b67b95522ab2d4b815ab1277245ff112 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.out_proj/_s.pt
index 81cb55ad68fc11947b8691dd272b63c2ec878bf8..5cb71a2faf26fa84d94eb8e76697be0a272a86ec 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.q_proj/_0.pt
index 25fae1ead109ad4c994a50d6a7af1a3e11f3e7ea..92cd12f053e35998e7b3f836bbf33071ad8884d1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.q_proj/_s.pt
index 9315a26245315a04e61d0e5ff5f5bc4a4afb7b7f..4af38dbd8653ca47add15fb43466356663f9897d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.v_proj/_0.pt
index 5a61943bd3e69b83ed4edc59108cc78dad0bfb7c..673f3bf73d616b5dcd26010176f4eea90accc980 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.v_proj/_s.pt
index 3f75d42ba4ca1862a718c28e4d2fa7d60a868280..9f081e32eb00797bac09d531897e1a82c5b87a39 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.12.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.fc1/_0.pt
index 317be616baaf110321a449f7571d7c3fbb2cac5a..63e03416706b01c98326d9920c18138e5e6b36b0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.fc1/_s.pt
index 319066a58e37715adb69b5187a04e9543d10bffe..556bc88ef8662f4dfa79dde9f4cc0a822b572b7d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.fc2/_0.pt
index b229b9ee48009455bf26dc6a913cfe1b42a8cc56..2222b53354243b15fce1512c9c1e715c3fb68c8a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.fc2/_s.pt
index df13907503a561e45494976e428844800ee87483..0979e48abf17df9458de46059921cac8781d78ea 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.k_proj/_0.pt
index 64ca7359c6b7f950a1fa00d9308a93c004bd08d2..8d70a0c2f068f8ebe31fc62c8d555161fae22eb8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.k_proj/_s.pt
index 575c075629c9c9c7d2834cbe48385cbcb018cd5a..541bcbd8f6ab3779a1d1e4951bbb3ea13bdd7a79 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.out_proj/_0.pt
index cc9791de5a68e1b14a388e0027fbf6fcd20a28c6..cb4ea4604a2e920583e1b7132c6a75a19647f63d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.out_proj/_s.pt
index e5f5ac432a2b524297845ea63b2e04b77ba705cc..eff2791a3fa1638d186ff9d86e347b3c7b1697fe 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.q_proj/_0.pt
index 0e21e9f80a7ae1dd469070bff4b9c836772577d6..13b5504ee0c4192a75b6b79ce455f0142d0a581d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.q_proj/_s.pt
index abe4e67ae313d0b47fc178e8cf552fe716f49442..01b18030c0eef881d7613d8c2f6c6a7557ebc716 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.v_proj/_0.pt
index 0a9f5e22cb0c7b6ceaa388ed5cd420893ba4a4bc..bccbc27b5eea70f52d75fd775ed4284ed7d086e1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.v_proj/_s.pt
index 8a0459871e624a812b1634fa3bbdcf019ad134e8..b2cf2550623ad62fe1088f9d5a633fa88038e229 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.13.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.fc1/_0.pt
index f671dbbd04f45ed3469c4ae9601d02e1c365287a..d191e49223041f43109db1a40cf3b61746a859f4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.fc1/_s.pt
index 1acc05850384b22720d5defcede99d887e1f2f4e..fa9c23d6d1bb619dd024c84be73f54c5629e003f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.fc2/_0.pt
index b572cfc14cb3cde8a030cb67acde7ee5c8d7f4d2..17a0be973576acbca9efccc75cbf1faeefc75b41 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.fc2/_s.pt
index cc2c43ffaa7e7dac95176342036eea386b8ea355..f30b712b4c1b26d50420616b0adb5629826d6943 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.k_proj/_0.pt
index 81ecce27906f1d7d774ad195e907b5f4bf6e525a..94e8bfd6be32f14456d5acb5ff4d6daa1ef6d3ed 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.k_proj/_s.pt
index af8e31c72ce09f447a885f025c3002845264f026..7ca57b69b17ee43ce27ca3c27db6b6122c1da846 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.out_proj/_0.pt
index 165bc7ab5d487d2e55e2bdac3ebad9546c2b0297..46edd0f016d63523551d334bc7e4f4ec02810a81 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.out_proj/_s.pt
index eab9a389191ddac78f59f714737da4d61d37fd57..356e247d6c83b1102586d41d2504d87c88008e84 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.q_proj/_0.pt
index b1704e5ddbc8420720d466c9476f14daad5fe021..b00d4edec7517071cebd7ed4fd79943648a75c1d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.q_proj/_s.pt
index 1112486c7ee613b1b197f790f723d74acafe10f0..a7863b08b656496a8a1fb7571883663e1a992648 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.v_proj/_0.pt
index 17d306fc2b6da498c50631a98c918a3e6d0c8ee1..acfe35b6bcab07bc004a7dec5862e8c57b97deef 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.v_proj/_s.pt
index 2da48f1ec37280e395da08167b51f1ad78adaf32..e0415182d433fb2d8ea4de4bcacd4c2b547cb969 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.14.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.fc1/_0.pt
index d17bd649cb8db7b9b29016bae4e2088b6e70fe23..a4cfaea3f4c36fd0bce1843a89b2d8e6289d740e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.fc1/_s.pt
index 05a29f076d34bd8079bed24d3d1ab0c98776a502..27fa43ef5f32a310269255b1fc7c3a4aade9e096 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.fc2/_0.pt
index a8b9981df02a07d5dfb8abd86022ad711cb2c4b4..f7d16c8ad9c63c8f7b2749cd12a98daf7f2f1734 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.fc2/_s.pt
index 88655bac90319ea5d9fb35d67af8c4fa39c62410..72df4f18a3f211c8fba207813d8303a51fd7cdf9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.k_proj/_0.pt
index 951a72efb0a93a768917d3e1f118296cd38849fb..dfc39ac7e1ec458d6a2e3243173b9269f178a2a8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.k_proj/_s.pt
index d314dfc889f51dc3ab85896c26fc7a516bb4be10..6907bfaddc7441f6da7609e246f5b28b5ba4e6eb 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.out_proj/_0.pt
index b52aa02b2186ab6f4102bb28b11dc42024d94b3d..672169d9708bfa84669917f5edc0232fb9f6d627 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.out_proj/_s.pt
index a003f5b89e841ef3c3123bba7332a26288b778e0..bc77be6705b5c67a4309a6915f2a50464564fdeb 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.q_proj/_0.pt
index be398c29cbcba1a89e898a2d4e0a1173b0e659a8..b384f64cbe2656d097cda67c7dbc70f036dc623f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.q_proj/_s.pt
index 3dd1856d0e712f528e874ef92628d253f8a415c9..eddecff2ecd2f4ac673fdb7539b0afd1c81ea093 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.v_proj/_0.pt
index 1ce67c73749a43f035b46d15c893ae5ada8191ac..564fcda20c143fce8699be43711b6d41a8caf660 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.v_proj/_s.pt
index 1e94c8a2fc979fcac48a57ffa008ef080798c61b..4b91c7501f77877ad33ab1b89b4c378c841e7539 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.15.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.fc1/_0.pt
index c20068cbe37f3729db8f30c8dcf05e07102720a5..c45f4c9f449e09ad8170519e9d9a1b33fb2041ec 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.fc1/_s.pt
index 1c2fe2c551b790ccdd40b57e861a5d1863d516da..822410c1cb9d4d30107a5fd7965fb4273ff7a78e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.fc2/_0.pt
index e6923eb158505d16830205ba011d1ff9057d7022..ab7a78b05e010bcc31bbbe94ccb8c6cf4351a9bc 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.fc2/_s.pt
index 1bca481ca1f057403d83b57fa15bf54394e6d0aa..cbc5a50a06c7ce132266bada2f3be423de01f035 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.k_proj/_0.pt
index 4a63074e12f1e99b89ac5823e00cf76a90831fcc..82d4415a08090fa15eb9f7ca40d6c7e7c9d0804e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.k_proj/_s.pt
index 111c364433e678ccc708619d03f2a0944e1c9d24..0ae6d4ce60b61aefb2f778de441a9ea404504e13 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.out_proj/_0.pt
index 718825e434113c0c0fa460e867ef419016b97603..2b8632f6352d923aa827fc9b95b09d2e9e36ffe3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.out_proj/_s.pt
index 049bb898b2b429aea897586f3272b81691c3f69b..c2ae84d1c9d079cd0dee9820a8c60d3e6914d34c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.q_proj/_0.pt
index f4ac3a8f747aad3cc8837e59f6f3f887cc888810..f6e31448d34a4fdd35e905316c4ae873120b71d7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.q_proj/_s.pt
index da3bdc27d028903995f0927b5b02fb947085c3f4..b2224706cc06752e12dd8ec047d747e021c0a1db 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.v_proj/_0.pt
index 78afd5ed7112a42b6ef1da4097057b4fe53ec884..e84c28e51cda0c032b1a55799f91cca1a39f60ae 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.v_proj/_s.pt
index f3e24c510e8d96fca59d7fc553f1972eb9fc3315..4fa7cc02ed314b227ada0caa1b610cb0440b930d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.16.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.fc1/_0.pt
index d335d5d21916f1ac0b0c67760c59e28dfeee48b7..3664fdd6adfd11c97300c8edba2b98e31097bc64 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.fc1/_s.pt
index e316562acbdd37b6d625a5816f166c3afa0be2c9..b05fbf99c27023ba5142c5b7f7b5d744e4884c99 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.fc2/_0.pt
index 7ee3c316bfae36312237d4cefb145509d187d1fc..3c1446149c69e1eb91e1e2256c4b0d13036244b3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.fc2/_s.pt
index d3a724ff773bde983a069c030ff16c807d2287d7..4f140ec7c8de13888217997de4c1bcd9be0afaa9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.k_proj/_0.pt
index 1ba64e59331748fad586ba219c9210e1de49003a..a54b41fdc011f8f61c30af5ed6712673cc57fc9a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.k_proj/_s.pt
index ded0864e21b01ecd6050649754b34799c98966e8..fa2c5b1f1cdbd6584f29775df709fc11e53a2efc 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.out_proj/_0.pt
index 077eaf0315f30e81f4561f02339030aaf22a9a6c..6bdbb040722503e9b74d66fc3f0bd5f81951f6e0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.out_proj/_s.pt
index 3fb68d0b9447b710d5e4fc2382cafb9fc2fb69d4..44f9b0da3a1dcaa6ccc9eccaf53be130641bf769 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.q_proj/_0.pt
index aa6ac51e6ed46851c00e315289b6efe90adccf95..7fcb8406af2139671e7ffe5f99c560030fe231c2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.q_proj/_s.pt
index 77b36485ad95ac14f968428da207b637f26352cb..e7726af383725041c7c5db3736102b183498f2ef 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.v_proj/_0.pt
index 2bd068eaff8096f2e776a9feaa56154cb316830d..668157d3c69b0620157a533d0d14514bc6a21b73 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.v_proj/_s.pt
index 7539ad60cfec2b276f08e2baf9cea3646aac5f81..62f1107adf4b82f952cf9cddc123f07300de2666 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.17.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.fc1/_0.pt
index 27313e0a4946781d9f6217cdb9b3123391d80d31..51b50b417a0d7b6aaa6595a2aa6ad1c90b85686c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.fc1/_s.pt
index a65e2fb3c6d8045627379ffa8907dc628da1865b..46b8766e1aa68909c11ec8ca409c1a4257cfd965 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.fc2/_0.pt
index 0c4c25fc43696bb14406b95a008421ab01ab66dc..514c0727362f340b833c08fc63872f5e289b3a28 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.fc2/_s.pt
index 3ae90c30fb0808b8e55d547cf4c0346c4cbd79a2..c06ae4c37a5c06ab23d8e592fd6682c8376714f3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.k_proj/_0.pt
index 7256c49e7dd111cf4bdece6ba5ffd691f67bc640..0d2ae0368ee84a3fef421c58ba2925b89413486c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.k_proj/_s.pt
index d62bc7c78f132331171e3146446d4585b54d4bcd..5e9c7589812f1a9605c8ed15aca817613c7928f5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.out_proj/_0.pt
index 3942d47b5c13d0159ceeae1b816842a56055bbc2..6dfa27830e44a5a48e0288e7469c231766d26755 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.out_proj/_s.pt
index 614e9c0d2e4dd1e72a5db81d68be60878ee851b0..de87c2e7aa7b69c352eec2376f688ed5836059c4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.q_proj/_0.pt
index 45e33ffada1dce8a939c5ce6af5b1cbfde2fe126..8d0f311c4b34ec035bcffdfd68187f633876cfc0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.q_proj/_s.pt
index 42e92c81994e75e6657ce9715c7362c4c5ddfd09..d8a9b674644b797f432a79b0bdcbc0b32ba2c828 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.v_proj/_0.pt
index ca95d8d19b648dd0fcc2c4f635b6af8aea1df8eb..35ed6b9d1e23a25c6f512045804eec79973eaa76 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.v_proj/_s.pt
index 0ef0d71c1d7baa54cca46c7f89b31d508d171434..b6e82ab3ff812e7a54b980d01cb40332f5a2cefa 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.18.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.fc1/_0.pt
index a41a531e721b02ae2849b06b17c72a5067f18d33..4fec85831c466f2a700d475ca35fa582c45b213f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.fc1/_s.pt
index 4972213dd4cbd6587cc53e00a0416e9637d26bbf..e815ab6d56ce57202746452e6b2d90513a76f526 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.fc2/_0.pt
index 058e7796cb5599771223b75681b8b9620b79170d..e4d203eb2a2786bc93c3c36cfd1ba2c37eff0af1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.fc2/_s.pt
index 3ba9ccebd7c0e1397f276eee538b2900db2c149d..28a3a28cca9150dee7c9ffb17f8bcc6b08d4c11d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.k_proj/_0.pt
index bd58a864cc08a33f2e0a0b1b08bbad42c007361a..619f46a30499ab572a52a003aea7b5d154aee941 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.k_proj/_s.pt
index 3008523b0e0e9497d8c76099aedefd14f4494e16..2c596f8427c1aeb7207e9055cb6177325e894da3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.out_proj/_0.pt
index 48c8ac8db4cf88fe2f4c565511be5ed60dd7423b..ff17bbcac23fe23aea53b69b9a51ab6df705817d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.out_proj/_s.pt
index 9b643f86855d2d177803f0dfc45af3827c1353d6..491dae4e62d966f034c33ed2e87a9634b030e26b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.q_proj/_0.pt
index c24a3e10af7ce77fc64ae3cea6236b663e78e808..ff9ff7c0a93bb645c82512561a57afa6726b0796 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.q_proj/_s.pt
index 9f8353729d8a1a9112c7b1bb5c1ce83153dc636f..a861bf92a42188ee5b3ad8427702e7aba35c3e35 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.v_proj/_0.pt
index 1cf70f6bb8bedeb27b9d930229d282c8f99fe144..0bd098f3475a012a832e30f04b36ca56c01b51c8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.v_proj/_s.pt
index 1b21d8fd7647660a8fd23de17187115ccc27992e..a5c03789382c9a3acdc33a59d640c1ce767b34e2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.19.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.fc1/_0.pt
index 548b74fd33dddeb609c3a9295621a0875206b156..c257e8d64a978d621bdc7dc302291129b86adeff 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.fc1/_s.pt
index 609b86305160646a0c61f2c9b7dd0d9360769842..62f3e2946a477fc46ca297c0308af6875950cb80 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.fc2/_0.pt
index 68e5d1cb31c359354d0131a18211fd79ad616c15..df4ebc787efa94e679811e6023219d8e7bec8fae 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.fc2/_s.pt
index e21865e4a6d05f8ea1fcb0c9b1a5bdf67ef5a94a..222374bc456606442d42c6e3ad61a8e019565eaf 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.k_proj/_0.pt
index 0fb6596422c817eb3d18232e6dc7738841343237..3e510ccf100f930573ca1469149769ef38d7b2f4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.k_proj/_s.pt
index c27969ee99d667dac79e8c48b63c81a04348a9e0..1caba4b35c94ac018c4dd4fc0da8acace40b7cbc 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.out_proj/_0.pt
index 972e517a99fde3e8f4a2a7d4c3062cd7ab903e99..b7cc514e16b2d39278e0712162205a32d6c61be6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.out_proj/_s.pt
index 8912e6a94fe6c4fe265cf29f1e652867fb0f10d6..8277807a857e069a373fa834f78a990b5bc651e4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.q_proj/_0.pt
index 739d600b37c27b852742826744ffc37ccdb391f0..ee634e7224342cd92fba00887fd60d69c65684d3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.q_proj/_s.pt
index e42b3125ca0bf6fac21b048b6fb96d3dd9cc3d68..94588f4f6254b11e81cdba481e54f263efd7c965 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.v_proj/_0.pt
index 9ff57c581d0fa8838e11838324c636c742e5073b..3526cb054475d3aa7fb2125fed339ba906a00eba 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.v_proj/_s.pt
index 9169b55113f13f99a703e3623bcd74f7a0bdc860..fb14aaedd8bbcd980c71de50d2a4c125d8d62bd4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.2.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.fc1/_0.pt
index 390f166b6a7ba5a56b1225eec4d95f5c98c11b2a..86b727c7c564b5a549b17984bf811081269b633a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.fc1/_s.pt
index db7c18ee0d455c669da0a906b36c24915b80db3a..aad0ef2db3588b1628429d41e26e05275a6e81a5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.fc2/_0.pt
index a51e1bdcfc9c6fde4f893400c69fc5d739263278..4058903a4f4a1552b24fb9d8a4760f814becf7ed 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.fc2/_s.pt
index 76356022fcdd4a8861f2155acfa7ab50f10be8f6..b04bf9b2b5a9aa11decb3602fb4ce115a2c9f521 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.k_proj/_0.pt
index 49df43f97ac161445c801b22be4f8576feed6af1..d31c01aed381cf581f2d20c1b8c55605307954ee 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.k_proj/_s.pt
index ae38abe548ed817504c012ee2daac5fe0ae0036e..e743aba61747966bdab02eea3f9ce4b8e0bd52ee 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.out_proj/_0.pt
index 3dc5b2a8bbc994bdd12d67257d135a94fd2dbd0d..4b8777da1e40138cd42866e145f17e741523891e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.out_proj/_s.pt
index 06d7c915d896c5e5eece0038a0c8833ff50400f5..0b72fcec8d2bd30e3f20bda515ffd258c41a9494 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.q_proj/_0.pt
index e46750701c671540192307e8ef7190c9f2fd162d..83b1d767b8f6c14a3aa6c4e6de8b27b999f3c831 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.q_proj/_s.pt
index c1de3ec46e414c53e475c5de7bd0a0a501d31a2f..d03c25c54a83c838564c2224ab1f42d96c71df49 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.v_proj/_0.pt
index aa8f635cacc607c2a863fe622ef95fa9eaaff79b..effa286696dbb9c7bf763f5a2aa9f324eb0e018a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.v_proj/_s.pt
index 4d2034a5eac27119520c44b2b62c0322bd773537..7c19fcf9cf6c929c9659f76dc6b003f8fd01c87d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.20.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.fc1/_0.pt
index e645b32b76b9a27a4fe92cb8d37f0cbe14f802de..52f44ba8813ee9aaeb44b10ddbc5f2462abb16dc 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.fc1/_s.pt
index cbe9f06cdc8acedcc9279016dbaba440e38abb73..dedc2fc31e81d17c8cb0fde1e75ce03752866d26 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.fc2/_0.pt
index ac7b06f7eea67c289db7a32aa4709627ed14abbe..846bf6caad1ad2b5a69c79c0b938a0db0782693c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.fc2/_s.pt
index a902913ee0fda32592015a2446455d470bb3bd2b..0a832a5dd12c3803e7b147b6a2fc4a0682546be1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.k_proj/_0.pt
index 8984eb047add6aa7cfda5cc589af112b18f1bcea..2edd0a46b14b90d96e4796aca22d1758d4dc1d7c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.k_proj/_s.pt
index 3d0f13f41cda79754f019ff91f1c4cb86ff85aae..71ea335dc62725b67d2d4db55e63a62012f409af 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.out_proj/_0.pt
index 25856ea5ab7fd36e41dd75b1b3ed969735dac5b6..12d08967c844dcee1e0f4e23319c848378265ac3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.out_proj/_s.pt
index 798877311cc24a8f6c3bfcd90ee8c53ec8b4f49d..9893d3ffcdf5eeef0db122045dc6a3fcbd9bda6b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.q_proj/_0.pt
index ad2f249fcee9d4779498413bd27767d63de90935..69517a5b3ae62e6f7c5f94d1f7530f14f10c61e1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.q_proj/_s.pt
index 7f8c0fbd03ddc81bdf4423a131d4072a49e46afe..c8ab414b0e981d258406643a7e7493582497b324 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.v_proj/_0.pt
index 2923c9cb51cd0b48b3292fa43c56af3cdcdac68c..a8ed997efcec7ae9581e7a7c889faf0f80e11972 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.v_proj/_s.pt
index ce5353f1622b43d73e25ead972bbcdf540d91c77..a31255e7531382a111b4bf703a4e6cc68e6bf084 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.21.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.fc1/_0.pt
index 61f243327ec6dc59f60e444887791dbddacfeedb..793d9951372d33c23b57cdeba9b90d7694e2d3b9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.fc1/_s.pt
index 932964687969d9e393597245cfa8dac587f31c80..2e1ffd46d40286a6d5dc270d0e8bdc15806e7f9c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.fc2/_0.pt
index fb7e50ce1066f500bf608f302cf35afc14a41d0a..933d8149661b4cc5ca17a24b98d352022ff915b2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.fc2/_s.pt
index 284c709593cf73ba9d075676c4bd17437bf4efbf..0653b4192a0055ac410f9a7a114fd5153a841ea3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.k_proj/_0.pt
index d9e1789770f26b903d3a84fb2a848d16d3d346f3..492ce243d40780072558060010a1353b582d5b13 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.k_proj/_s.pt
index df42ff3659f38fa28b1bfe2f0c17c0576b6473ea..29018ed2116ea8b81f1463bcccbefe9b33c9efc6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.out_proj/_0.pt
index 6b8c02fb19623b6a43875602e6eed1b199232823..2a7c6b8a414c02e4527cf5372a6cf6da8a9ddcd8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.out_proj/_s.pt
index 322e5283333cf5aa30402a44fcb7504513d5aabb..fa4ff95a0e51da129774f2643d298f38fd0b707f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.q_proj/_0.pt
index 29b5851037d61c15fa640aad911f3d2433fb1b24..041ba610666040a40dceb2478b672e272ce1625f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.q_proj/_s.pt
index f9d75839170a6e5cb38a1cbe6f320cc98d5b7db3..cb4d3d33bd0497893755de451535c536dc184d3c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.v_proj/_0.pt
index 092c1ddbbc39fc90734a79460f5f80c332212523..a05709198e40c1ea330cf96f67a448e6e0825d46 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.v_proj/_s.pt
index 500800cecbc30963d8c8a116231a05a3cd9fa686..76643380c742e1514ad0f574934b52eace4d39cd 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.22.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.fc1/_0.pt
index 70c0d4d9fec4100e69effa1acdfb6f2cdcc0ab4c..0eca5adce500d993d0663042d0113c529c8b7d60 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.fc1/_s.pt
index b734b25f4255016b6d4a4cb127639236a565b38e..f1648d6001acb516d1e7fd8ff70d99baf4405f9a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.fc2/_0.pt
index 4c80379ac379e6cb6645a0fc47f8cc5f9db6a5d3..223e65bb4c50f1fc72d6f6444ce52b72d1d03210 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.fc2/_s.pt
index b2162d6dcb123001ae39ac118f3b5cc4cd7cced7..a891800133958bc846d3ab7e1fa7baa58438c9b6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.k_proj/_0.pt
index 8c057e51d8325277c574fea69bb5d854c5ebc440..9d70c48e8f1fcb1e19ed1c787a0f1df64db06eb6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.k_proj/_s.pt
index e562086bcac869c2c66a89faad48651a386435a6..bebc3b7634cfdac5cb194ec4a21cf6d2d8a8ddba 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.out_proj/_0.pt
index 77f4e66889b16edc2e1b61885abfeace5350408a..76f6cd53dacba08f90286e54ce4efe778070c251 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.out_proj/_s.pt
index 0ec202a4606f62592206341deb7cfea3fb824e1d..f5e34375badbac9d6c63547bb844725421dcbc6e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.q_proj/_0.pt
index b31b3fdac3b8f263b34c223cd90a9decaf098a12..6cda7843a6158ccc99a8675ce4e09deed5bf9e57 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.q_proj/_s.pt
index 0c54c31f26873521b8535181e82196500c679d67..bd35c9e38734cada072d3927dc5b0e9d9e227c91 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.v_proj/_0.pt
index c9825d5378799bd641498aca87f153b065ef636f..11d07fe9bf6fb123db8fca1255168a33503c4ad2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.v_proj/_s.pt
index 8db9e75b0290068ea5a94ad7a62545d0588df0e6..d2a9ebe8e187f328f8b35d815d186aa0c77feebc 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.23.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.fc1/_0.pt
index 6d181b6ba78dd133f2e2747b6703da25b06764e0..aa2310006042381025fd105d3f742e6ed20a4ca1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.fc1/_s.pt
index a3adfe92005045af47b814c9f9577874209ff07d..725b751ce75cfc8249bb708b8bd14343200c1121 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.fc2/_0.pt
index cec03d117f2c98d4e50c6c5b0d28e409e710e95f..0c263567e80dda221ac6658f83bee48ed39ab529 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.fc2/_s.pt
index 38256a105cd576fe0abff773cdfad440ca299683..35fd6cbc67ff645eb6326d796e442c1f8e1a4ab9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.k_proj/_0.pt
index 469cb6925585c914b57c13398be53dc60526ec7d..ed418f9658b78a9134ce7a111535997d3633e2c6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.k_proj/_s.pt
index d575b970aec3f282123bbcd537bbe3799cb12c34..533335cb7328a57ad91e42f4744f184144f93dd2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.out_proj/_0.pt
index 487ecd27efd7a750095edb9687a94dd3174d8c02..0d379797edbdd90bf4f89430484f64d30c895857 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.out_proj/_s.pt
index bc9c76d426af1931df7e16cfa475b498e7a21cda..cca847b13a63c6c10aa667afd2391a198ce56685 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.q_proj/_0.pt
index fbb14d27fe0eb42179ee7b186ae6245df04a2bab..2b9be7529f97d7e497d87f4c04b6951b4c9680ab 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.q_proj/_s.pt
index 1094b8418a1321a0e63434592529995c9195f068..87436a032a220320084f5ffdb7f6f9aad747a1d0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.v_proj/_0.pt
index 8d884c65b44e5627867525ee798ca719278ef92a..4f2ac1ec3e2ad7a14e8feaf3744dd4bf38aab85d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.v_proj/_s.pt
index 29a2a7eabeafdb59d920c3f352e8bcacf4dd3e65..30d214876afdb93b4d24057129218b0ad57ec5a6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.3.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.fc1/_0.pt
index 3844ba1b39cf7858c7b1f037fd235135badfd974..78459f8f071f2d2a06b7f9597aefdce58dfad781 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.fc1/_s.pt
index 5df20526b5bfb1fa70479161649b7def8a774aa4..51ed68b4633ecbfa8188cbabaae26c442a0c1578 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.fc2/_0.pt
index 980a38faed06acc1f75a9103dc5d61e67e64cb8d..41113ab47ee8473d7bd0aa84b0b99ccc3c0e9a13 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.fc2/_s.pt
index 0423974af1d2ba730d7e0c062933355d588e2825..e7e9729a2a96863bca116d88eddd411ffc19f3f9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.k_proj/_0.pt
index 616de995c62c39c30054f4f3b9eea93a8e77dc8a..ecd54373cac632716252e3e873f3e8344d455a9a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.k_proj/_s.pt
index a3143acf8b0f36593eacf8c13ebda77073182a65..a1ac53a9a317d118ce9f86e367c9a02982b10306 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.out_proj/_0.pt
index 169525b24e6a93e685f26f7b6b4be0474f7a7d6b..f1c260440b9bd12ee337ddc095958082f40bbbc9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.out_proj/_s.pt
index dda9d1805ecb772d803d7d6f6234ca076bc07d15..7b307b15902c0fea078af84ac0552313fb3f59a0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.q_proj/_0.pt
index 10ba2fb48af3397540451e3306b8f83a26e96965..8a2510fd65f184a2bf32712a23980057fd787727 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.q_proj/_s.pt
index 96cfbeac1e174d8385b8a359afdc7a05ae226491..1bfe2198f13b55be14e7eb532a68ba1fc8c31790 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.v_proj/_0.pt
index 37a2dd64aa5318ad049364f72b359d187ed21ac5..c96afdec8e3b42e0231c5350ceea2003b11739a0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.v_proj/_s.pt
index 439d9790725f1c5372e97f2220a6f09cff3936c6..668b64bda13e4bbc03e5ebd4a56d459b9dcef52a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.4.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.fc1/_0.pt
index 298e08b57a98eee7335f13fdaf8847c1a4a514a4..d3e7c4a32ae11e662534c0b1a187f8371fe03c2b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.fc1/_s.pt
index d4177bc9c5456dd657bf61a1d9a47c72961a71ad..497dcda656d54ccac0cb8ea521e08417bd1dacab 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.fc2/_0.pt
index 5e01d7b01447f428fd88af5716c7782d6bc01c9b..0fb4ea10082f8ce7a0e2cc6fc95e581ef588fb06 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.fc2/_s.pt
index 5d1ee2f367c508ce1c7a40078182aebeb7f1af42..c554bbe9fa0303248b8491d016de4ad5bbe67a78 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.k_proj/_0.pt
index c2557a824761146222cd8b688206b65ef59376e5..7de5339f489c20be30eb0c1a348c88c040543608 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.k_proj/_s.pt
index 22e341d6edb7040e344446e13794d730b60a0323..7b0e094c10ebe6a3d0b81c17131b4c5cca2ca9e7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.out_proj/_0.pt
index 2024eb577834ab53349a3e46de7e1b7f0da643f6..6a89500466b2615e36543e974d90874e00481420 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.out_proj/_s.pt
index 0c90fc9c92c958c3a7b0fdaee662aa3b75f3870e..d6d29c346b6aa479a0884c909849dc1273d6c9e1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.q_proj/_0.pt
index d95cdb4ee7259d0b22f83cb6b456894b88d78355..187ce331cfdc4a042b797f75f6e31be2fad10cb7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.q_proj/_s.pt
index 0877f74ced905e4f015571c2930a56cfb9d28654..6bf8ca610bcf156d5313934700a63503d673e5e0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.v_proj/_0.pt
index 291233fe20eb6a04b3006d841c14da4ad076e2c2..8b8d4759ed3d2e70c90a15916e105b5d5f53680e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.v_proj/_s.pt
index 48b54b7a3f08e5199b55946a053f3d7f44dd9b71..c96dc981f121df0d0845d5111ced5bdcae98e3fe 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.5.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.fc1/_0.pt
index c8a798d1e410bf821eca5889fba881ce76a3b6ec..75bc3c97433e7bbfd0bf2c87b55e94046e94cdd4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.fc1/_s.pt
index e1b379f2126fcb41ee93d9c453749a530fa736cd..7851aaef7502bc0593b97df16b41dbee2db78135 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.fc2/_0.pt
index 3af612400a786553f7264568901800491c50d889..be00a852222e2f77f1412fa6ba6e090d4387a100 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.fc2/_s.pt
index cf52b73c18e29b4e1b0f78b9c41c0a42c0f4035b..9ab475576206fd66ccb22e0b02057f2ca0521e58 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.k_proj/_0.pt
index 8c894b3c20d27a50f7a8e6b8f0d2274636a132bd..924e67cd46f9b0f122e375cc538ef851addead70 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.k_proj/_s.pt
index f7e4373049e176a5e6f65280c34bc1b8c68a1147..5ab4a16cffd52d4d5c6b28754eff656250c41289 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.out_proj/_0.pt
index 223e275a30f251647d2c9ef7de4f81caf1cb62ed..2d55fce105964f6b6090c7f1a90480dd5beb3430 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.out_proj/_s.pt
index 40bdfaa6b60862f5ab4b2fb0c4cc0f645e587552..20d87fcee301be475d4a989324d3f2a6a50e6dd7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.q_proj/_0.pt
index d9576431a9bde603ecef935848a27bcc3e97d733..1cd6273fc864b9ccb9fb0fea951a541f04e00c9c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.q_proj/_s.pt
index 89c9fd2d1db57ee99fea3eda0d65c0372f7d4b37..b9c88b90f911f5855874abc348566c611b2bf4ae 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.v_proj/_0.pt
index f33510b3cdd829b47876a1ade61e292b4febcf1b..720b9b215a456160622aa5404aec509452b4a775 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.v_proj/_s.pt
index ed59e595621b26cc260c4d35b774feb38c89ce3e..c6bd8e892fe9093c4834a47f6d0ba94d84fb27a9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.6.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.fc1/_0.pt
index ef7dac0100b412fbcdbaab39ffa4fac62faf6929..0e8e3be6c0529312c6de0a5f82d158b09caf55f4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.fc1/_s.pt
index b1df93ed115ad5bfa61182d344e85451342b7674..14846fc1eecd4dd2f767d6b147930b9f73dc8f75 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.fc2/_0.pt
index 37ac1a29adeca290590e894753cc9fd7f0dc4cf3..4a55d9529dae8cdbdccc6ebe6555484cdfc83779 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.fc2/_s.pt
index 64ecb3ef58c6bda5f60fcedf88d04f334ad74082..0b5fb2fb0ae2fa5e8a46df3cdaa21700d4c4d8a1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.k_proj/_0.pt
index 2ff60f718c030561de081b8dbfc2592b6e506f84..0221eebb8ab27d307c31204a7c04ae24c2bc2ee9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.k_proj/_s.pt
index 76050d876dcac4878b45f3eb2017e874cf841bda..aa4a2755fca53fbe80b77ad0ed8032306ab79f13 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.out_proj/_0.pt
index 0dd60da07fd837f1d271cb06ac2b98e414a64801..348ea9878f6a40c9dfbf2d0ceca2dfa1f3f4f563 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.out_proj/_s.pt
index f3d085f5b8232e841152fab77d613539a8221c0e..8fc574335e5c84f481e895eb1fa21b90b480c428 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.q_proj/_0.pt
index 03e191f5a80ef7843eac0f530550daddb4896479..b0a13e18b05d8141c8491abd92a27226fdedac63 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.q_proj/_s.pt
index 7efb67869060985ea47d7354d41147e55565234e..2ae282b5f67948e41c9ffbb5a2476534ebb75a63 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.v_proj/_0.pt
index d0583939874b54f029ec80d019b9f4359005ba34..406e4300a94353cca25c7057ed96452edb05f162 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.v_proj/_s.pt
index 40f1d339936c2722bb1fe9b3457a9d00ce933242..97c18d52cb64e0aaf1a09193a5d6941da2d13452 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.7.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.fc1/_0.pt
index e77db01a9ac2d47056689731c4be59211b4cba0a..e780bb015eec88f457aed399439e7f2763111581 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.fc1/_s.pt
index 3289189309fb29ccc312150fd9882ff7e4f23670..c898587f138bdf78e860b6e49e8ed549ae850f08 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.fc2/_0.pt
index efe4e9a674aa8e3cce96b5065e063fc9c75b3802..9ca393197978fbf9a5493d41cce40661c353e217 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.fc2/_s.pt
index 9397bc44ebb730483bb8655d5e4ae6e939e71dfb..2e2c27c2a334f0e93df4b0af42eebea6c70e75e5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.k_proj/_0.pt
index eedba70a9879d36f7e24eaadbc9895da39a2796d..647b2744cb85c4aa3d60906ff70251638daa938a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.k_proj/_s.pt
index 36b0c4a844c20c4930b94bc913897095bc74be1a..a1a4a9f5074d8cd1cf64528f24b434cb7f2e8ce7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.out_proj/_0.pt
index 6f4ebd436c32c2f7031790abee959545fe32755a..265d186beb966763bef7482e61bba40137165df9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.out_proj/_s.pt
index 09bfef9c13efbb7220c625875806ac2d851f29cf..b2180040a867494c20297e71f5fcafff28163d08 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.q_proj/_0.pt
index f003965e6ba250ccbd21e5f1d67eccd855df8260..72e9b80f243b6417db767f15675fff723e3cba91 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.q_proj/_s.pt
index d6941cb13cc5764302aed1aa7b6d7bfe21f66f65..f116109048a4183d4018ae1417553ee54b664b1e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.v_proj/_0.pt
index ebf64f3525e5e83a03eaa398dfb157ec5086b375..69a44afe40b3c99445227fb0c4d7631c594487d6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.v_proj/_s.pt
index 8c6fcc53b44b553a39d3344a6dfdbdb0f18bc29f..0abd088eb4ffa5bdf84227f1e2aa8263c93af5cd 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.8.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.fc1/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.fc1/_0.pt
index f963f1402628f39615db675b23bb4de21b4d4b6b..22f84a2d0f3ce60a686825b1f8c27d4bb09d64ce 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.fc1/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.fc1/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.fc1/_s.pt
index 763609ca425221e3d754348ccbb33e48ce741dd8..43b1640ffe8769b09d2962f83060c589ddaf8f68 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.fc2/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.fc2/_0.pt
index 79505bdcae75eabb3e2463435cfffc96791e6de2..e84e04544f6e22f54b9a326fa2b165fcf65da086 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.fc2/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.fc2/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.fc2/_s.pt
index 6d9ef87ff50db16380359ff385fe4e40c597b4d8..c666311ae88a1f84faf505dea35a75dafa9b21c1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.k_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.k_proj/_0.pt
index 2f443b7f59a393d0f2f6bf4bcaf683b1c736e929..edc16130c587563aecc741d820b0ae85c2b7e597 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.k_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.k_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.k_proj/_s.pt
index 797f5e86c80963ca3c450fb0b15dfd8e8073af8e..e32ba3149bf9164448160def0480860b581f19a2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.out_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.out_proj/_0.pt
index bf269bfcbcba58d80620b09d939619ba7ecc9004..045fa9837e285e25a2730e9aeffae60accf2de92 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.out_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.out_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.out_proj/_s.pt
index 326cc93d1424c72566f304b8bc438859573bf11e..36b2aa56d3cfcd2dbe7817f6e680de9cae5ea3eb 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.q_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.q_proj/_0.pt
index 71efe1abb3480762c1d8ee86ae00c86ec80e908b..386ac55dbf6234d15cd5fc24e2d465d5d7cc5e47 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.q_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.q_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.q_proj/_s.pt
index d875115ce3e13135220d7550e6286ce2f6649fed..9164cdf269220c763b1328ec6fbffbfeab3be816 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.v_proj/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.v_proj/_0.pt
index 2080a72537131fd9fe1b574e6f13b7288d3f738b..c05af830a82c52bec1444798f01e1c9999a1c193 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.v_proj/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.v_proj/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.v_proj/_s.pt
index 036d1f08346d412638769a4a620cde9af50aec38..8ace91f8b294df8289b52ca52baef8f1f54e11cb 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.layers.9.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.project_in/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.project_in/_0.pt
index 0e48ed011df873ecd666e52d262ffa013956204e..bb9b25bc378755a22b0645c5e0f758fb74cd1ba8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.project_in/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.project_in/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.project_in/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.project_in/_s.pt
index 8d0f5cf354bc5d7dd32926eb0c861ca60f168b98..2e4354bfe5064e08bc9589b32aa2cc0cc1acd87f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.project_in/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.project_in/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.project_out/_0.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.project_out/_0.pt
index 0ce8ff7ab546118d1f0772015b9e0b795f31251b..a17c403873646aca65cec1ad81b6eba128bedb32 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.project_out/_0.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.project_out/_0.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.project_out/_s.pt b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.project_out/_s.pt
index 17033ff5098ac85fb8ebca49c2d4d32c05f1f623..c4e55d5c30e39fdd20eecef5c3cbcf7f076326f3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.project_out/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_1/trained/model.decoder.project_out/_s.pt differ