diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/lm_head/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/lm_head/_s.pt
index 4b389ed5b859b60538bcd655b3da7ada642d3f8d..378899ab5f04bceb27f716ba8ef3d2a076cd4834 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/lm_head/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/lm_head/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.fc1/_s.pt
index a1115b63c4533eaf79ba6ab5bef2ef5ee70acfcc..8d5b982df6a61de923abcb2180b450cc8e151fae 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.fc2/_s.pt
index 869b4ef0f83d9938b041804cb78a8d019c369976..7047c39fe7dd46123525abe031ea4b1268effbd6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.self_attn.k_proj/_s.pt
index 3c152e238357c353879cc627371d0ff3a36d20e2..4f3cae6444bef26e28acfcf2f9e337a108efd864 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.self_attn.out_proj/_s.pt
index e8064c75ab192341835d38e58501d11b7d4465b1..0b678ebeefd812b202f9031ea3c345573d794441 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.self_attn.q_proj/_s.pt
index e29c4b6f4a7fa33914f4381ec42ec404c8a301ec..357e25ed1bab0bff1409365041a6f8ae980c7c85 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.self_attn.v_proj/_s.pt
index 96319ee3974100d913533fb30a32d8d470d34c1d..95f331d01804b3b7d8495e0500a6adc7b802130d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.0.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.fc1/_s.pt
index 5ef55ed1ce5e0cfb33583cdad366d0aae5b74fb5..bd5853351dd83ac5079b755c893a2e471139e766 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.fc2/_s.pt
index 4ccacbe19d222a96c0d94632a77a2b4e7bd5a7bd..384d52ff6fee560a0a6edf7f98ea9c869d7ed179 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.self_attn.k_proj/_s.pt
index b18b712cfadd7c610b4ef9df5bdbd74863b3fa3e..fdc7160972aaacbe492fed80d6c4915e2382351f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.self_attn.out_proj/_s.pt
index 6a29fd6488e1019e5a3a5fd080aab5ed6ce3dc68..530a290ef6d8bc0bbcce19120faecffdf41f234d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.self_attn.q_proj/_s.pt
index 51357aafd2cb39a31217a732d18f980de5f5e0b3..6b8a7b28be042ac5e9b91cdddbc46ac991f9efa6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.self_attn.v_proj/_s.pt
index 939f58ee87f91ca3a648e956320b89f8cebcf830..954e5d0d122b9ed843051130d617bd8c3e53a4c5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.1.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.fc1/_s.pt
index 8b22348ee86639c143657792de99ed6a8b3747b0..f18d4bb5c876be16a531e049f248978bc0611a06 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.fc2/_s.pt
index 706bb0369410039b23a9f56dbc49fec5cdf49f6e..9a5a1b9d842a80cafe8b3309e48b27e4ee7b3c28 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.self_attn.k_proj/_s.pt
index 586a01839fc0c99cee7e85dff4402ea464cd3258..b5e07140aa8eff2e6156acab346b22d1ef68a073 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.self_attn.out_proj/_s.pt
index 2d5a719e0af3c787035274d804cdcfa85bfa796a..d6803fc4ffdf899893f38aaf23e93a2ef741d935 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.self_attn.q_proj/_s.pt
index 542690f7b98519dd549cc780d35078bd1a350f8a..34f645813508eceb78adf3d882cc21ac1a886ad2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.self_attn.v_proj/_s.pt
index ca0c8ba60059272703c6415063572ce43f7ad463..6d9f88f3072d065d4add8376eab84785bb136bf5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.10.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.fc1/_s.pt
index e6d225eb053bafb9325657031455aaa695dba2ab..28aaa205cd843cc353fdaa9ffe990cfd45bc11f7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.fc2/_s.pt
index ead3abedd8b47bb1ab5b9200a1d813863de2af47..5c1eadeb1b2047b3c616539e7c5e9ce733e888ec 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.self_attn.k_proj/_s.pt
index 8241cbbabf7815a22c0646f7028738e4b1340ef8..bd33d383effcfb74cffd954681e7c0e67ff19281 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.self_attn.out_proj/_s.pt
index fe01025d6c8cf232e5c33d81edf79b0ebef85b19..81671d2a24342517e551ccf451d4aaf6411ce0a3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.self_attn.q_proj/_s.pt
index 8c489cebfe93ff3514a67d409429cf656a8c4edc..3384c992ade049e002d5772c343724548d214311 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.self_attn.v_proj/_s.pt
index 6a9db598b656f8e817adb0fe1752240b60222a50..366528c7c2bbc34a1cb2c5fb121621d26243bc45 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.11.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.fc1/_s.pt
index c332b18fbc2f953d94d3ad65120d0cd8086dfde3..7b08a9678d577c76105210a5cf2409f63e58911c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.fc2/_s.pt
index 9847b7abead6a87b013354e4e9b2e7e891649a12..e8c1d5ed06f30461ad5d255d27c05781694d10ba 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.self_attn.k_proj/_s.pt
index f0816e1b7a6c16e538ff55a9a54cbb394e0a19ad..00b2b3ee179e346afa6bf2fbeda0e1bced05931b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.self_attn.out_proj/_s.pt
index 541eccdd3197fb5e396fe2813136d701c7fdcf13..9781f8a6ac11679875a302413d264f89c0ae7e7a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.self_attn.q_proj/_s.pt
index ef6e7cd1b889997d913f97a6af47c4d0ef6a1a84..52205637460d1a82c4415cce1f1317642c171a17 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.self_attn.v_proj/_s.pt
index 8d1902ce1af01a790f264f85bbaf569405f73ccb..e7689b9b618b7667ffb18b18f17525ceb07cddb8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.12.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.fc1/_s.pt
index aec93e9e8f6f5213b3b2b2f4e163c4ab8b3308dd..34ede6ada17eb4458c569927916ea5c90b8f8fcf 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.fc2/_s.pt
index 3064ba4dfe1e36a73a60762c7115e49b2d1b19c1..d43a7eedc4baed4ec0a8f1b0cfd08d8f16622a7a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.self_attn.k_proj/_s.pt
index 0a7d6376b3d0cf9349da6fb4dde1eab3762706c7..c90b923397ad71c2ab61def71492a6a84c09a9e9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.self_attn.out_proj/_s.pt
index af0128fa7b2bae2a2a356ba92c9486b186c4cfdd..5e156fe89b47f5bafc1418e0f351565568c9876e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.self_attn.q_proj/_s.pt
index 822edfe926237461e65f31b871138b35ccdd16bc..1e3c3c854bc2c5236657c95796454ce22ee2242c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.self_attn.v_proj/_s.pt
index 881932501d47c32f918741349064afda8f1208fc..94c6261bb6d3f7e859887a60ef320b274d24c57b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.13.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.fc1/_s.pt
index 9c960ba604fb93d7df7e0f5232af617847a75b75..035c4ae031dafc10e7e6fb7e2342bf77f9806c39 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.fc2/_s.pt
index 69ad2f2d159ec75cbd957178f41ab7c44ed49dcd..4f37941e881f316c45eb16de16250057e6912d08 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.self_attn.k_proj/_s.pt
index 306cae7f422e4d0194e54ae3441cf86aa152096b..23e8a0ba002f56fca553487182830b49dca11975 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.self_attn.out_proj/_s.pt
index e8bf8443707cdb499147df67d7889d47665b9c16..ac295b516768f95439819a79a3bf3bf67f4d013b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.self_attn.q_proj/_s.pt
index 13d90aea231e3e225d00a6b512c6428cf37a4046..ac0d454157c4d7f3a7d734116cdd9a39589692b7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.self_attn.v_proj/_s.pt
index f4ed1ab552a1e2d30a050c7a484d0b06c56937ce..d3e399bc61700262153f3bdf229316e3af4b0de7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.14.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.fc1/_s.pt
index cc2aee85450e4f40446073f920bf4f78099ee3ce..3e44fbddb55ce55232bbed4db791eb86eccfc7e8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.fc2/_s.pt
index 777d983437e929b3267af51fd48ab42f0dead7b5..057426c2d6c9c3118f5cbc52e94af6a82317637a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.self_attn.k_proj/_s.pt
index 3b58b1eb864dada0c408985267bbab5c260c3a4b..9def4e5584dc62f132fbc34c410f2a90da79b91c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.self_attn.out_proj/_s.pt
index fa2b1822a0ae58a9929216028e8c45d9cf02f7fc..129778d9d1196709861347ee49d293b316873424 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.self_attn.q_proj/_s.pt
index 858baa3c68911d15755b574b0cf7e711c03a7f6b..f9aa4ba9b0c939ae1cc2d4c66b6bd3659df4fd27 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.self_attn.v_proj/_s.pt
index c1e42cf54443c5ebcc149741fd8614c63eac9a69..0ff0e375af1cbfcd4257b4bdf8abd3accb4f4ea1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.15.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.fc1/_s.pt
index 40dd7c4869d14ce64f91179be7af6b0a9b53c4a1..36e7c1a335f3db6869a2a8cc6a4d104f50b2b6aa 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.fc2/_s.pt
index 7a0e28de0d5a834168781bde0b296c715d400c2f..7aa9f5a08f26fcb3715701b89c4405ab717ce5b9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.self_attn.k_proj/_s.pt
index 0f78b1eeee402c59e960dad8cecb2685f345e983..bace8ecc71b3d6ba024f20dda12f49852b9c37e3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.self_attn.out_proj/_s.pt
index 4fb227bfa373648b0c431d47d29d16ed2bce717e..628b323820ff1cd159bf1d34faacdb7c76a3585e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.self_attn.q_proj/_s.pt
index 248822ac83f8a493bbf61aa09238c7a02f6872bf..65824be0a475df441cf078668ec3cd17d771aa62 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.self_attn.v_proj/_s.pt
index ab3b264a6928f572ff13818da64b94d1b82d3862..b0b6bea5d75ff47a3dfeed5929174be2efe40fd4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.16.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.fc1/_s.pt
index 2c6ed00e91967d2eac6476980d33591c3ab307c8..d0a4178c09d97c5e88ae5f08705dfb7689cddc38 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.fc2/_s.pt
index 60de621857f364f7ca74c1445185e1f960850dca..39acc7617689085cd7554d6cbc7deb033f37044a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.self_attn.k_proj/_s.pt
index aa85480c8fa4eaa2156d5432b41fdb328258b72d..056a7e8e3a6e513e33c0e51ee683e3ccef954184 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.self_attn.out_proj/_s.pt
index 605ef5e559f563a233f95255f8e3ce9d6d31ba3e..c7d95e4ffd4421109f72b92d274253e8534c61d8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.self_attn.q_proj/_s.pt
index b827ca7b877f0f8434a83a9705ce0cf8d4514409..3559ecd803eb83e7964fd9672ca1a15a5514b92a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.self_attn.v_proj/_s.pt
index 3e49e2139f6ce52e9af4ad1d0c75d18e324503fd..7e82824bc22c4ffc5506510a97d04cef51a06081 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.17.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.fc1/_s.pt
index 13778f129973a3fc1f40e3fc889478a5bb3f6cdd..10415e69e90d95e93776056418274e7fde4a93fd 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.fc2/_s.pt
index 2b905f09592e77c19ec8ef58e74dcae30e84ea58..080fb188460800b9fd7f7479f905da59a9b9b608 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.self_attn.k_proj/_s.pt
index 5ba1180731f0c2b1b5efc5955d500a17c60fd67a..24224c9ef6c4b9d7789b53a3aa177d18c3c1ad02 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.self_attn.out_proj/_s.pt
index 0151b7fde3bd6825a1dd89877ae12f8c7cf7d9c7..2ce7a6e4cfccb0151fad7296df93eed355b0f709 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.self_attn.q_proj/_s.pt
index c786b7147d4c35aad9178872d0fa17e725b384f9..c7920feed31b366756ed8f040533b645c4e97994 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.self_attn.v_proj/_s.pt
index 920ad7f4f40faeaa5c05556001746e6fe44eef6c..e206689bda690ebfbbc43196d35f9a06534319f6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.18.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.fc1/_s.pt
index 6bb0dff203b71ba9fe73ea07d73f241c5f182e4a..2487e9e55b4643536a7e17de424f7ad7fb55e90e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.fc2/_s.pt
index 89e143de34be080bbab090e51cdc1db366deda73..2614430adcc23f8e42a015c8f5c68e8eb541ce32 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.self_attn.k_proj/_s.pt
index d63fcb7fddd96ba76eeaa3176684cc4454279b59..4621b51bb476db12078a585f2795c94ed817820d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.self_attn.out_proj/_s.pt
index 4dff301007c43087dcc33067f732209f72654da3..6d1f05d041b1fdcfcf7f9b25f406fb43dc1a76f3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.self_attn.q_proj/_s.pt
index 988246d1dd5c1515f815b854b1e0bdcbec5bd31a..9fbd9dfeb9abb4c67bf5198589226d8f195cc2ec 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.self_attn.v_proj/_s.pt
index 08c6f9019b83572ff6f62036dbec85e3c9a30774..4d2992c56b20a0cf9bc1d4f8e39181db3498551f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.19.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.fc1/_s.pt
index 5c530b4c64e649b7fb2cb2a1c2cc0a0074404040..425abb0f14933413fd6bce41feecf48b7a8af791 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.fc2/_s.pt
index 5afa36ccc7be46e2baff3723d71843553ec4a5fe..28bac0f25bc0adfab01a80a866258d6022daf0d7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.self_attn.k_proj/_s.pt
index c02c2d49ebd01aa28e7f61276f7356b08163ab42..cc0261da1eb15bc3f3339fadd679fe54fee17b6e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.self_attn.out_proj/_s.pt
index 1c61371dd2e8db037af3afc84b699c3d6f3c6390..ee826845f10cc26453fce398b575d60aaae44bcf 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.self_attn.q_proj/_s.pt
index e885d5c62aa9a88608805feaf4ee53f7997be782..c8eeb897fdec14116e85915bd3c300cafb4c47ff 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.self_attn.v_proj/_s.pt
index 35d3f2f2711dafe976a82d96463ff3b17ed51302..67f01fee7e0b8eaba7d0366b3c00a9c3d163c7fe 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.2.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.fc1/_s.pt
index d06c5858780cffbe60dc460366829933ad5b5f7a..f6ee11450a5fbd96c0fd0c02d676922bfd60e12d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.fc2/_s.pt
index 89a72a736a6afd3a0d0dea36eb50b8a888c218e6..825f9fead40e0a4e92876ddf207ca382ab4d84c9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.self_attn.k_proj/_s.pt
index a7d6a1fbfe493dfc136f36b6e9c4da2e8c7ca800..30c0cfa66055ac5f26f6a283406127476c197227 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.self_attn.out_proj/_s.pt
index fa2ad8a1c1cf35fe760cf0fb8a985af84fd90bfc..550c020ea8904e65057bac19248737bfb60e950d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.self_attn.q_proj/_s.pt
index 94e96f6d9a0848ccabac16911fd26966b00a0415..134c1bee271b28e93209a8c8e323dcc00069f884 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.self_attn.v_proj/_s.pt
index 91aacffc6f24cb41ec43bbe00182b84a912c8788..c444c2865bc56e93a8e31520e9fa4a96716d3228 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.20.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.fc1/_s.pt
index bad00af1c6503339a6b393cddc7071eb32e0ae5b..5051ec1b31fa00f35c7bbf2ac38ff705b08b1e97 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.fc2/_s.pt
index 2dd044429fe3d518b76fe9e67cfc3009f0a8501e..7acde02ac57905d8c375da5078b007e31992f785 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.self_attn.k_proj/_s.pt
index 033b6bd9d65cadde45f456e23f8fad547d0a40d6..c04c973b0bb5459b02aa0c944872a45352076b03 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.self_attn.out_proj/_s.pt
index f236c0fbed0375610f9e9b2fcfbb817f7dfac793..5f33ba85c92a98f97974dc555bc876a77445378b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.self_attn.q_proj/_s.pt
index 2482c40919e1debcf19bf1d0385e9697ad39526f..08f17bcbb2886abc429f478a4c02e21027a85f9d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.self_attn.v_proj/_s.pt
index 35c8ac19cf0a422e45ca381c7fa0af42264d2ea5..77ace04ac37f117f7f355934b7f7abef5ebaa1f8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.21.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.fc1/_s.pt
index cc01d4716657d7f041f2795b7f929b52f2a188b5..9f352d0c288eaec304737d5d8420bea94425049a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.fc2/_s.pt
index fd0ee1d0681f9dd38c00b70df5911f7b023e31a7..02a406602d9db43255bb713f8de494d48a56a360 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.self_attn.k_proj/_s.pt
index ccb6d82b088cee1974ff5b84d63ccbdbe63080a0..9db2d509404482e448ffa0f9af98946bf352c468 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.self_attn.out_proj/_s.pt
index 774194e3459965fcc84fa0158b8385aabe6951e5..8ca076d71b8d9d316404f1cc565c5d1b25d34e11 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.self_attn.q_proj/_s.pt
index 3a0eb5cf8823f9797ce6942cbb5e7662ca28a1aa..6049e8964ed609101562bd16bfd15ff16b46dea9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.self_attn.v_proj/_s.pt
index 7c77ccd208fe31f210a0fa99c6eccb50295e13ac..5b97868cfb14d810b976a4dbdfabde4b86ea935b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.22.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.fc1/_s.pt
index 99359f0c68743546f548ace94ba2c994a143d8a7..9c53f44a91366b7509fc380b9e82b63e389f7156 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.fc2/_s.pt
index f8970e8cbd8631540968928fcfff444d640698ae..040bf475a3ce8ffaa0f2b70b433d6e81e8ae1638 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.self_attn.k_proj/_s.pt
index 291ae8ff1fbe6a5366c14c7787624adf146a39db..298e7c9ade1a20b821bb928c550bcb3e5cd4cd05 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.self_attn.out_proj/_s.pt
index 9e8b841c260a522d060f568e7de4484de41d312f..f99d13fd032d065a3abb914b972ca798a900fb3e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.self_attn.q_proj/_s.pt
index 13d4d33170833be5a4088f919cb7733d0787541c..95d43ae9d25462ce0a8765994b59285aef4e9335 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.self_attn.v_proj/_s.pt
index 3e29ef03fcdd148d2263b94b1373bbb88336a153..d081ec1c9c08a8218ec746422fee637f10d3eb8b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.23.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.fc1/_s.pt
index ab6818cf80296afdab8606513b79f8cc3cea4874..6ab7c538f03406b65758941f1055246d762cacbb 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.fc2/_s.pt
index bdd33dab3efc556961a2c0b80483e5232594913b..fc77de905fb83a40bf9b321b8278d626033fb54a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.self_attn.k_proj/_s.pt
index 447bc147729015b61cb68795d7ff1bb3c534f4ef..c95f5ceadc67e0404ffe2dc5e0de4c68699ccf4d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.self_attn.out_proj/_s.pt
index 611ed6c8a35c5148f85b77a159f27e500e9a9c73..b75bedcaf24f5e6e3d70ba047b3c1a119d87bd7b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.self_attn.q_proj/_s.pt
index cf3f6bd44aa0a04ab460a2d196de4f44112c282e..e8dc4f9786f417efc8f0eae7db9d0be0e0312039 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.self_attn.v_proj/_s.pt
index b80073ce9eca6e4223937780c6319ad36d7eb941..19237266f574c7f307a1574744d750847d0668c5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.3.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.fc1/_s.pt
index 6d2497d289775079bb92665c3846bb1ab4b443d7..0cc511306660717407318a1db6b799c247c788b0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.fc2/_s.pt
index e5cb16b5155dafbf9daaa2fe011b3b2fbc3596d3..a9eb81f14b39413a70f62eda24c4527990f9f105 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.self_attn.k_proj/_s.pt
index c9868b1f925f29566621c664cfce42b0a2ab61bd..ffe44a085fe4b311e86434453737ed5cebe5104c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.self_attn.out_proj/_s.pt
index 3754c81f6d1e62c8bd97c18fc07665d42835f7c6..a20e2c064d838e001b47c71fdd512d9302dda61e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.self_attn.q_proj/_s.pt
index e3a7380d364d3575827c71fc5f3782c1306cc3e5..9face6b6a4ab633ad6270a3d9b94d8ad414b2aab 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.self_attn.v_proj/_s.pt
index 5ca0202c0863dbda0c84088af3fb5dc93581fc82..a4d664e3c3f74ee24c35669d83a74ab75152db00 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.4.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.fc1/_s.pt
index 65116b8c8e509e00dc33c0f7aed19be1070d2780..f74b1f4474ecdb9a546ac45ba846feb0b299cf17 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.fc2/_s.pt
index f0aef1e9de36546a60b12c65a16f01474f43f9ed..f4d7c16f257dae5897b9f0a41ba12f91659fcc46 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.self_attn.k_proj/_s.pt
index 07c3124305932ccc7e9530512bfd8cd0bfde7682..fc66012abd33a664f4bab3fbaa31a89da6a33c4a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.self_attn.out_proj/_s.pt
index 2be13708126d6f6a44c1d78747c48925cc43855c..d162ca92e4cb4ba6d7ea15582e2fb828f6aa4df0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.self_attn.q_proj/_s.pt
index 43cd37af6343c2e050a5dee9eb6696aca58afa6e..a6f46ebfb129f9a2029bd00bef129ae372e8ee98 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.self_attn.v_proj/_s.pt
index 53f95c1a2450567d88c01a41cc48c0a0629c1adb..8d87ac48179d22f2832c1c02ec9f606bd6d487d0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.5.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.fc1/_s.pt
index cb7db40179ae91b77183c9296f5c1789a3794e01..5f2f11555442e6b053e63544b35adb0b7791e6dd 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.fc2/_s.pt
index d41b228287ff30787290c86b3e7abf18b22e3c20..ce394cf1767451be64587a3499e5c3c0d77600a8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.self_attn.k_proj/_s.pt
index f2330b64d12d6fe42fa2da920efbcdc031ba1d4c..4cf0882b7c4b3a50f83b111ce49ff8767a2be9d4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.self_attn.out_proj/_s.pt
index 8cdc7edcca5a0781227ac60a4fd3110d3043d26c..80aa6bd9b2c55411f2846fa389c2b78966a9f434 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.self_attn.q_proj/_s.pt
index 57ab750cc990ac584d656f571b8dad0694cc1993..1b964db56dac952355566601a06bb9c0e17a1cf6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.self_attn.v_proj/_s.pt
index 1c06a9a8ed792ed4fc06716f8cf85c0e91577d16..061f9b9bf2d3f4e60c52ab6461edd64eccbc781c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.6.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.fc1/_s.pt
index 9f747df4dbd14350785ab6e89f7eef6e5335e186..944e232f57be1d0eb123a4ee38a4014a87deac85 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.fc2/_s.pt
index 5d757db761f0566c192f795744ef45cb16a12935..1c69c9c38d91adf40438aa13a4562692e2e724d1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.self_attn.k_proj/_s.pt
index 503f985624ac6d560bd9329c942751125887c62c..1fc13fe36443d0cd5d47f9808b822a4fcd027236 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.self_attn.out_proj/_s.pt
index a586354b2033ddb7f1c0fd63ee32e5751869139e..32a2a374f5ae480652fde468cd2ae5289cc48f36 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.self_attn.q_proj/_s.pt
index 214e29d6c2e1e16aa1fcf85e518fb4d6ac26b58e..52a24a9ca546a5027bdea1977c7ca5edcf0d35f9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.self_attn.v_proj/_s.pt
index ce19be6c6fb4380215ded63892c2cf6308415762..09f38e9d4bf3da36b6ff13f2d2e8b00ee6906b65 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.7.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.fc1/_s.pt
index 2e9fa0c784af923e0c0703c251a32c609778e3f5..399daedc2da1cc4912064baf54ce28e2eee07ab7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.fc2/_s.pt
index fc04d33ea628d078b7916f6db9e767d0459d834d..c6d551b317cd8f9e10b0815d49bdff5f9bf017e2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.self_attn.k_proj/_s.pt
index ae7d6f6cc0771d42973eb9d240853fae4566f5c3..c218f2ee59cc88135a8f000ccd9f061b547c94b0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.self_attn.out_proj/_s.pt
index fb86a0971b976bd5f8acfcffc30c2948a7915e63..953d82931c5b87ff34f2d45db1632a262c491a32 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.self_attn.q_proj/_s.pt
index 81edd80449d09dbcfef35edb7779b49060d33284..457fee5de8f23ddb6d9d922fadb7fd22085c496d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.self_attn.v_proj/_s.pt
index 8dc0a0c34dcd2f14dda265b2558dc0cffea6e98f..9d952b1cbd9ac578be318401c56b036d8ce64174 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.8.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.fc1/_s.pt
index 03933cd5bf217c9719ee3ce14df3dcbe1d979aab..3525a0fb7fc42f40e4fe442603af64d8dd1539f2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.fc2/_s.pt
index 562a9cfdcf5cda6cd88c636217a762f1278e6283..cf5d9d7373218e4a30d5f2efb9ad83dde529d419 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.self_attn.k_proj/_s.pt
index 6727a8f1ffea5735c57c9e354297f0efe59e6e78..d4cb2a5ee74bf9bd2d44908498d1915705b5f6d2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.self_attn.out_proj/_s.pt
index d6ab9c37e57d57808a37bfa618bbcce0ff446345..f87ddb12d8000eefae8e6c07292f41442c6b522f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.self_attn.q_proj/_s.pt
index 66c03771609431d22a9b25cbf3416fa70f67b3cd..89a5c28471f07f076b4d72fe211a29bbd0ddfa96 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.self_attn.v_proj/_s.pt
index 9a38984f7c48c370d5399c9f8d1e41c6629da643..835fa5fbd4c7f935eb40e64162cc58b83b29fa12 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.layers.9.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.project_in/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.project_in/_s.pt
index eb28560ccabf20a8017587417c1f36e79cc3bd5a..2522eea5dc0eb5528cd95e95e41c5a79d24fc343 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.project_in/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.project_in/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.project_out/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.project_out/_s.pt
index 534c595cda3ed1c5230d14b50f5f3c8c61b6ce60..06a41c50466c0f3e37c53d2389adb5f0d1af7649 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.project_out/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/init/model.decoder.project_out/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/lm_head/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/lm_head/_s.pt
index 4776fd14ff082f7b11db40823a74b92e88b79211..2a458c64a8e823c8745d52989ee640e21a1b6e87 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/lm_head/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/lm_head/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.fc1/_s.pt
index ca0dc08bb0f4637529cbc7a5c394e1220749ce84..bfce497c63d022f4e7a3f7f763c91134f3bd25b5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.fc2/_s.pt
index 0f9e34d4a595bac43f9b15e82789a83738e604d7..505355cf47dc0058dc58103c7b82082edda51b48 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.self_attn.k_proj/_s.pt
index ed5b16efd6a8dfc4b1bd8d146b785305520234ae..280fab2deec492167fe35af91fa4ce08e88afecd 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.self_attn.out_proj/_s.pt
index 0aa783b871937159fc49ee31183eb8f4050ec648..f1acb52d8aa910108c6ace0b3a6466fd64ea4eb4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.self_attn.q_proj/_s.pt
index e082e0c8464fa740f465914c5e8fe74007d4c91f..ae16f0c2ca88aa0efa0a86ad48ae2e80bc1664fc 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.self_attn.v_proj/_s.pt
index c02d73b7af01816604f6348e25b384cfc771075b..07ce84102d9e28b4e99352ff603fcd4ecc93490f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.0.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.fc1/_s.pt
index 44c77d469c7cb7209c0e772461595c35e0a28598..d10ae5c5e7a21edbacf9292c21f1cd708a44bb52 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.fc2/_s.pt
index 4f59f1cf411a8102a5bb782b173f4cb2e78ad57d..288c276e2bf04db6c694b27c117133279b5e272e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.self_attn.k_proj/_s.pt
index 76f288d3fd0b135a80383bbeb6a8570c89dbf88e..8455a315d5041559db3be98126244a0e4f723857 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.self_attn.out_proj/_s.pt
index 7c987eeacb0bb671a1d93549e9e463d197dc5ccb..dd971144ce9f7dace4672200dddaa712a4462205 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.self_attn.q_proj/_s.pt
index d2702fb47a27453ae056a24c5ce3d6b28d657e63..12fe8d998cb168497cf41b9d703d5fca1a0dc087 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.self_attn.v_proj/_s.pt
index 4a7808e3fc81c595eb4740b2d9c339602d7d100c..e639c0b1d6f00fa37c42fca9fef702d52f7234c4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.1.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.fc1/_s.pt
index 9fe9bb012d7aecec4235d0e1d417fbc39bb2b85e..8770c3fa72dd0a1c42436a17c4a7cda49de63e0d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.fc2/_s.pt
index 9a9d59f9bb277eafb873e2a4efe03bc0c9612d79..fb2ed989585036b040cf2ccf747957d9059c028b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.self_attn.k_proj/_s.pt
index f70bb482a3dbe6e759d6e26bb2a1360b0bf84ba1..d826bd2bdfc255e4bd7a1e5f70319d1b04d9f3e7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.self_attn.out_proj/_s.pt
index fb92271917926ffd55c3f162325e406edfcc62e3..083e6eccff23674df296095aa99f0b8b85d940b9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.self_attn.q_proj/_s.pt
index 1543102c56938817bfa633ca36dac41b8f71ab5c..278f7985f76066f9c09f61cf5eb3653c58d8fdfb 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.self_attn.v_proj/_s.pt
index 593ad50adae04806a4864cde3a5da4beb48111d8..2030d27f806fb9c7aa82318354d0441601850401 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.10.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.fc1/_s.pt
index e3611c55fa4acaec15a0cb39b8e52442954320f9..bcc6a26c413ab105acae3956a0479c1b3a7bad0b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.fc2/_s.pt
index bbf4e986252e7e428b14d78d276405d06abf9d07..f15b80f9b2cfc637cd4cb5a76bdbcdab75aded08 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.self_attn.k_proj/_s.pt
index b24c45f23e6bf2c39f5ef58e63a272cc0c7e053d..bff8cad7c2403f9c94a438550762b6f66f739f2b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.self_attn.out_proj/_s.pt
index e0884c18b398b5112368bcd7cf1d56f249ed4224..d4e7478c90fbbb6d32821920ed13ca027caabc9d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.self_attn.q_proj/_s.pt
index be0f4407ddfce6b694880cff47a43dae61c8a4a2..099ade4ef0781ccdf7b4a7f212e24a35b489614b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.self_attn.v_proj/_s.pt
index 70508620b19607b10dc07afe7b33f18e733f0df2..fc84be94a82dcb950ced342da268359d619802e7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.11.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.fc1/_s.pt
index 41a17050782a3ce3b01e0f12db6ea1260b3b2c07..a903379d36ae361f45669569e8d14c32fae5a0e6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.fc2/_s.pt
index df969ce5eb9aece420aacbb3f88257b8d7905ff8..634280935b78ea2ca63aa6e6698bbffb80f6892f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.self_attn.k_proj/_s.pt
index ee768db9ffcb12f89f6a18649e7aa2982b8ed825..eb683b21ffba451aaca74728b4ff658376725180 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.self_attn.out_proj/_s.pt
index c625919566384b8937c903ef8720d34c17187f8a..a893a75b077165d2a5dd8a714e052bce5ea90dd3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.self_attn.q_proj/_s.pt
index 70583b0422c49d23719bdd2a9c21fc5b6bf70353..c00807516895245afeaf2a8a4ba0cebe8f14d750 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.self_attn.v_proj/_s.pt
index 9b6b8dd0429b8726718a4e53d205b7219a6127fa..b136b8fc7bc95ab39936db5b8fb1f750bd845d28 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.12.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.fc1/_s.pt
index 2b4b186c42769ba8d881344a82b8543cefb0ed7e..540342974c42d5658fa44025a9f7cc99893cd5af 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.fc2/_s.pt
index 5da894e0c4174ecf1fc5de5205245431bcf88b52..4e557f63419a8e9af5a786d0b297c901c68ff8b9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.self_attn.k_proj/_s.pt
index 205fa18703abb2941a340ead7626564c77e98b94..fe2b1f8496690249f88754128a0b45ecfeb0dfa5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.self_attn.out_proj/_s.pt
index 047bfc7f8310074104efd51276a2a967cf57e97c..e8e8ca50a5bc733014733948487a583db431f959 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.self_attn.q_proj/_s.pt
index f51c6e88d6d2df09b7febc0e133daa6040a49fb4..649d02c79a93d0ec9209a910a6351b1094b8f72c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.self_attn.v_proj/_s.pt
index f7537ff39c6638e8f5c3400950bd6c81869e6420..6eb4538dabdf471c6319ed5f4058fe67613de177 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.13.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.fc1/_s.pt
index e78bbe73c430b7f06c3eb06bbc71e6c167f3c9ed..62bfa474cf070355587c59fe804c909d99092636 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.fc2/_s.pt
index 4ad25b0d4a1d289ec82dfdf56cb9172ce48c5c65..f4136f02103391a15a985143f712a75e76066e98 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.self_attn.k_proj/_s.pt
index a72115fa0d556c7183483d486fe8e5512f323429..5ee0c2fc86c8362896827d36b877535bd42f1cc3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.self_attn.out_proj/_s.pt
index 8eb6284763987806f8ab7b155a8e1bccc9ffbc11..b2852fefb150672b5be672d7accd524a810d521d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.self_attn.q_proj/_s.pt
index cc947c5666ee1626bfe52bbe9fce750f5afa8579..b22ec365688225458ad5ff9d6a4fc7506ce71fc2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.self_attn.v_proj/_s.pt
index 36712999e23c269d78f00e90cca8da51abcded21..953319817ff8a62d8d47bc3f7293af018ec185f8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.14.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.fc1/_s.pt
index b01021069995876e31b605eccf670c81039493d9..defa9132acf50bffbaf679cdf7bd2636104e5158 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.fc2/_s.pt
index 7f3932177e0d6c911e953b7a52e088e78a9fd4a9..504d8ad120dbd45db5b56a35370590a2af615e43 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.self_attn.k_proj/_s.pt
index 58d4a27c6853cd5951f9d2f93fe25b54349a5ea1..daa64d211af7de863974854e6fa11f850cc94a05 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.self_attn.out_proj/_s.pt
index 086d2f1b7435aa2e9baedd79c6c1ed0715033c0e..e65760e61fd5a46f0cdf68c98a59bfa3efe98f1d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.self_attn.q_proj/_s.pt
index a32cd8e4077724a122b13763e51f62a13b00387f..04ff4d61ad12363f278334de67225851e2429ba8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.self_attn.v_proj/_s.pt
index 3c0d68326c8adbedae270bd574bf0fc61a8fbeea..1ed98b7d94fe6b130556084c88e597d9dd3747bc 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.15.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.fc1/_s.pt
index 7787f426af419d735481bf7c8a2e756ee858e011..23383d0fedb5114a78b71231ce897111f243f8a7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.fc2/_s.pt
index 2331e1c8e0f870f10d3020344594267a3a7cd4a0..4da4f2e079ac7d80c3a9b221d3f050652b6f0b41 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.self_attn.k_proj/_s.pt
index 56d9f4301fe6116cb0631de5b48cacc802ecc46d..aca0cbba2b690e7658edaf4562857f7df9d38447 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.self_attn.out_proj/_s.pt
index 2e057bc8a3d297b09411ffa3e67a00cf8d2bef2f..1f0cf514833b629a9b31160e049c5b309f354c3b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.self_attn.q_proj/_s.pt
index 80ad596b7b3f3d2c011f7fdf09dde0742ae7ba06..a574806f0408773bb7b44d12810cfc047dc102e0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.self_attn.v_proj/_s.pt
index a674242a155cce073a73c2f168a7b57f575816f1..bdf196f2e7a7f2efe740c95743a38caa0a564b1c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.16.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.fc1/_s.pt
index 13d1382db9554772bc94b21abac6c78b09b34203..67097e64b4fc636e872560256ce4ab2d30d23c16 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.fc2/_s.pt
index 158c56c6df1f8f3639daae6d5859a4bc28b33f59..148b15541d1726aee039e38d0f894c04bdc46747 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.self_attn.k_proj/_s.pt
index 1671e5276d5e333f697fdef89b00035c9b32c666..9b17fc9e6c418858d52e496109881391d801719a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.self_attn.out_proj/_s.pt
index 79eea28049f590690ad04947de1ade7b9c796de9..0f15ddbfd97b214b34f45d2aa5c5fd932f99efd2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.self_attn.q_proj/_s.pt
index 21b28fe7ec62317e7b78998f735e3ba76a15f9e3..03b1fc91dd5e20d354d3cc712de48618406d709c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.self_attn.v_proj/_s.pt
index 82e9559544afe43995dabb618a746f264c5c2516..087591e3ea4e71689bcd3d37d2a7c843e4a952e4 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.17.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.fc1/_s.pt
index b48aff77d067a550238bb4cb3d6049534be9dd34..12279ed632408e5227f51aec7ecd7d1566e4fe99 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.fc2/_s.pt
index 4719c87a8870198393fb78f0c67fee3f0b1c7124..9414e1e38f45d86ca1444b3d792a4e23d6f946ad 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.self_attn.k_proj/_s.pt
index ad5e55a7ab3d897c6f4bdb4838faddd106546765..37b1f303da355a3af230ef60fb70ae4b5515249e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.self_attn.out_proj/_s.pt
index 19c616c7c64b8213c5bfb5f1832dd62c17b77a76..9eca136d2142668df5169e2cb2d7399c058a388b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.self_attn.q_proj/_s.pt
index c841f890cf82e0ed406280b0a0695387cbe840dd..73003cbf28dfc42e3b64ecfb90ede87de040188a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.self_attn.v_proj/_s.pt
index be54afe51891010017b9d8be092cc8757b616cb8..b0044f245bdc5cddff5ec6da386b767b22a9b489 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.18.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.fc1/_s.pt
index 4648916309271ae71e476ea084914e9908956c96..5b0df1abfcf1aa60cc24194c8e6558f375340c1d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.fc2/_s.pt
index e88b126d8a003ad47083faca8c4e2623e4102d88..32c1bbe51349229bb500fb9f84978949da067bf7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.self_attn.k_proj/_s.pt
index 32741e69b56ec1a74066d80d6d14789c33afc49a..6d0fc6f457e8be738b9cd05b19c2faeeabea5b47 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.self_attn.out_proj/_s.pt
index 2eeb91d3da687788d9d41e8f1e71326c78fdda0c..0f61a6f95c1f92b1628b575c7590ed185ced49b6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.self_attn.q_proj/_s.pt
index 83a7f79588619c0508c3eca73eae41d6323f3437..79d3d1202855659eca94ce80ea1b7a1ec1a03273 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.self_attn.v_proj/_s.pt
index 19bea350a756bc65b0e89322a13e1e31a74ef84a..85c42f1b2521de85a2bf2eb435803a03ed9eba4b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.19.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.fc1/_s.pt
index 6953d9f754f8516c9e8f84bdba99585a07e1a4e4..216c86bb48f6bade18932dcc161d18146ad3239e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.fc2/_s.pt
index 252c660dc8a961eb0c25a35df6b183d64c994b22..3dab920d6b43b35e9af0b9e6455afdf5590208c8 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.self_attn.k_proj/_s.pt
index 2e701cd3aa48dec0b9951d67e5938be67c2c8535..185bf67cfdcd225d71507328c122395d0c79756e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.self_attn.out_proj/_s.pt
index 46c6c1b187bf7f0c21e9f6d57e4f03f34839cce0..10c9b17e16d3f3f7f7f4e990abb93e3b86ad04ad 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.self_attn.q_proj/_s.pt
index d1911476bba330eb5a93225f19e9a89f0c43805d..83cd74652b330a685a7e30e06719c1eb2ddf73c5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.self_attn.v_proj/_s.pt
index 6facf4b4b413eb721b5578695a133f493ce8fdcd..e9e2f96ab7ea72d8eeb2b43d4bfe56e703446587 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.2.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.fc1/_s.pt
index 9591335a36cf668f7b760fc666631f8e3cf0bad2..9e96e65bfe073f8ab6d95fbd2aafc6e6df71c9a7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.fc2/_s.pt
index 6cd24560857bef9856636de35253323f90c57cff..7a3644578a8e69554d451bdea39fc621bb814d02 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.self_attn.k_proj/_s.pt
index 445e89c5ff10681daad7bed61169cdaad1605ce5..013be3d09917fb089325dcb2dc001674d1f9bbc5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.self_attn.out_proj/_s.pt
index ea3f0821a8177b7fa5adce5562d1b437823348b7..b7546bb5344338dd1747f652c40bd20b22bec664 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.self_attn.q_proj/_s.pt
index 0a811363d15d47464aff0e5a1b23a6fe82ebf57d..7462449c5f260cd51b2ad5192c92f60b69ddcc19 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.self_attn.v_proj/_s.pt
index 5a9bd3c293649ad85a7e361ef7c8eb2213c01dbc..9849b43055c7977ab53af003c2bbb55bdd6cddda 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.20.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.fc1/_s.pt
index 16c05a191cf6679e52b149aab29fd084ff6bc6bd..6f83955e456d34f8fd11f6678bfcc2157b16c809 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.fc2/_s.pt
index f853fc7ed7a576a8407fed3c28ad48a90780ee96..a5912bb7d4c4b206437c5c1b2419b1cec5fc91c2 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.self_attn.k_proj/_s.pt
index 0a68110981caa7f604dfcb7b53233369493f936a..bf3daa25e2548dfd490fcf0e9afbee6f7b06a35e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.self_attn.out_proj/_s.pt
index c934ed6f8068a43391ef146aa6d16b3214709c44..90405ac476454d3ecc2db35dfcfb6658f15f5794 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.self_attn.q_proj/_s.pt
index 923bbe416254d8642496d0e8f345c15b591de405..592bedc9e58ff27b0808b48a18c9dd6dda5e435b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.self_attn.v_proj/_s.pt
index 2c0a0f3359f0b01e0185f0ffbb25dc253f5a862b..9ecaadfc19e50c9ad4ad300600e97d4bb335a80c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.21.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.fc1/_s.pt
index 0a887d8d7c605193a37a58191fb97687c21c88f8..7dfe5e4fe9771f54f4989cbc1453dc3fa78fdcd0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.fc2/_s.pt
index 8104ac876bcd739af867bbdb3ca2398c11a626fe..c80b2d954181ae30a09f9678cd8dfa51a4961edf 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.self_attn.k_proj/_s.pt
index fd4608e3d9e848b5d617d3b66a9639f7b6951f31..c2d821b5efe3e877ba973c7d8bf05a8012bfc69e 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.self_attn.out_proj/_s.pt
index 7fe604ecbfde8883e2972911f51a15002b211111..010fcfaba6e6927f96b95a296d4c9ed7b0d6e532 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.self_attn.q_proj/_s.pt
index 0d9a32698ee0684ba5eeab29bf3d4cb938f286ab..d8b33f30252e33a3afca3ce82b8df4dd7c7a88d0 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.self_attn.v_proj/_s.pt
index f0b5bc668f57cee00d218df449d8e973baca2336..3a0db86a07baa7bd3084418d37e159b9531ac0a7 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.22.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.fc1/_s.pt
index 1932ee84586425bfac01c7e194c5224289b7222b..ff51e32ce84aa8cf688bf4b799e92eb28449e532 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.fc2/_s.pt
index 8f179454cd2ebd3ad67eaf45ca3913c25e7c4885..cdaf7a3effc015daa034b8024c158f483fa3c0a5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.self_attn.k_proj/_s.pt
index e35675c8ae97c1ff4a4771b645d1d6e1890150c5..8312c9aee8e218404f77a109f31c37a51611376f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.self_attn.out_proj/_s.pt
index 5bd105bad3680acb53dd4f6cea2ddd9ee379fcf9..10082f6a52ec1cfa23de5c18bb2920d7055afa6f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.self_attn.q_proj/_s.pt
index be2b9fc7e589d9a4d3cc1c503907a914f68f08f5..bed53fbeeb9d103b3860c8c408eae06c584c9eca 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.self_attn.v_proj/_s.pt
index 8c6e3aea4cb87be508426c13e9fab6c94f244f69..fc2c638da8bffcef79b487bc58ea482b24158c2d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.23.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.fc1/_s.pt
index b7ed8e0398a0d4e62c6ac7db5227a3fcb6d982e4..027d2e0236262bad4734990a1ded3d71eecb10e5 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.fc2/_s.pt
index 0008a95f89300c88d1bef793977f32771b39cba7..992287492a6789c76ef11231919c9b699009f0ce 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.self_attn.k_proj/_s.pt
index 32bb4fb8b3c8cf75aa2e39ddd0a6843423255f27..723003100de45c475bbbb11b083a78b2305d2fab 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.self_attn.out_proj/_s.pt
index 6bacad70a9582068879f9063f912cd764753fca5..4bedd597073456f7a839aadcc76aafad0fc37796 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.self_attn.q_proj/_s.pt
index 09bcb439380ecd261174295a8beb78cf8c993506..9a5d32bf63a4c160d9f9933a1a2d489713cacc51 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.self_attn.v_proj/_s.pt
index a862c235db43da3854c1442571e7fb34c94105ab..3f8f944ada1993bdf79649b8b8dae338e7d95745 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.3.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.fc1/_s.pt
index b0deeb5acb096c2b527497d8bf97f16720222abc..425b90ebd7484eae00d376ea1b756a27b0e3898f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.fc2/_s.pt
index 751e8a3bf525dc3d50f861777eaa8db40543d341..84a24aa7ad206b699342eb74638a82cb3bfc7dfb 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.self_attn.k_proj/_s.pt
index cb2dc0fe3c5aa3071e42419220f7ce120d81ba54..7ef4902e837e8d4e55c719f819592ff35c860732 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.self_attn.out_proj/_s.pt
index e0c4e5b632bac2cfe885066528595bc92e88696d..be13ad7e9c05d5d755de1ba891d92e30ee1ea4df 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.self_attn.q_proj/_s.pt
index d26137ab23d37a76510f30b60cf755afb9d5fa09..8b1118462be4c33a2f114be0ab1f9d129fa2546a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.self_attn.v_proj/_s.pt
index a06a8339be7a2adf4e7838ee284a043b66dc424c..d7be3196a5b8386b7f1a66a3d9bb5ddeece25b25 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.4.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.fc1/_s.pt
index ac65367c818f4d646df831f3c210a35bc4cf6a55..6df98bc455fea60a9659cf29ddb9c991505db9e1 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.fc2/_s.pt
index 841d719d3eb73172d42c868ee28bd00d1331d264..b2f5a92d56df379d22123e7026a8a0c6d2e71730 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.self_attn.k_proj/_s.pt
index 2001949500350cfb33514ff924a9b4acb9e425e6..f2c04a4b6e8386ac519e023bc3ce877379119aa9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.self_attn.out_proj/_s.pt
index 0e7c16e2b293b3c16d1dfe63b3daea69dbee4f31..a0a3fad0833a0a721ed214d5fa0099a0311ba05c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.self_attn.q_proj/_s.pt
index b357d5a1595f64778eeb0069305a069f310e7fbd..82680c3773dc94ab231ac187dabb1cc781166e9c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.self_attn.v_proj/_s.pt
index 81c38e3fad1d7341c93cb2fb340f46fdc5cad61b..5d4eeae32930515fad3621eb095dd30122f28741 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.5.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.fc1/_s.pt
index f1312ec5351081708103717160e91efc9bcadae7..a786e1321e42d225f794aa4bbf64bd02cac7635d 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.fc2/_s.pt
index fbef4e88a7e8044f73246fde9f21804fcb0427f0..89135d915d0bf4a361058955541bd1dfa2b52580 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.self_attn.k_proj/_s.pt
index e93f7f83cf4c5edef90e6d83cdb8e1c6bb955345..08ac4eabe53aac9c9e0ccb785fdcca15eb1439d6 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.self_attn.out_proj/_s.pt
index af5c7ebac1ad47aaef4b8c35eed6ae038c2267a6..f11e18bdadcced15ca9b3df2ace3ba77c37fd9d3 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.self_attn.q_proj/_s.pt
index eb723e1381b99433b4463105f88a5a16307b75c0..23cfa26f1fca2f0a8c8d71b93bff49e98c85ddf9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.self_attn.v_proj/_s.pt
index de683f0aa1ad71d912dc3db66289962d3a86bda4..868710b8943521ce8341a7a169bad52ac7302454 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.6.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.fc1/_s.pt
index 596e872e7229724d41264df5d454b87a5d68b9b4..a120204ba6b837413a2ad9d69b67a1214b870c90 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.fc2/_s.pt
index 627e7cda7ab236ad766cbac0245c039fd25eff20..55265f63cb32e378a9fdef8f55795cc07d210250 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.self_attn.k_proj/_s.pt
index 45c174cd9d40eb18eec2634527d9a7b7a1283d85..2904f88242a2a121e284709a0fe50512ea2c6bc9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.self_attn.out_proj/_s.pt
index 37b3e756155d1a77a5fad225c3e3f9a593f7c658..a1b7496e42d2effd7b1df215b0c13c05d1950a24 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.self_attn.q_proj/_s.pt
index 96953d7f8fe0c37afc4e04469a68a76911d09f53..0aee337e0c3472bdd0dda0c99946b01b56d1345f 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.self_attn.v_proj/_s.pt
index 7e70290be38da9e458d0b713ebcdcf588d6893a2..d90d1f7935049da6f37fb2baba38d3acc1e19a4a 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.7.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.fc1/_s.pt
index 1f98dcb9f282467a0f6319c1d79af0a13abf3887..4711b740fa01c8da6303cfd2f1ad683eba8d25cc 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.fc2/_s.pt
index 9857ba1021dca1872a0a9b139ee8ce794cd94651..11feda66018f0835e46b6bc3eca59a0072b5743b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.self_attn.k_proj/_s.pt
index ea8b90aba6c136d22005560e97266e59c649e1b5..327589f7ad40cfb6deb069f93dd8c023b27b382b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.self_attn.out_proj/_s.pt
index 1e02b95fb0e64674a05d0d480979055acd47390e..a965572dc2ee6948dee96d1015f729aa5cef702b 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.self_attn.q_proj/_s.pt
index ddad8ec311e4eabf574efc3db744a37d1745a239..da5b8f8fc32c85f3dd5749b64022e2852592fae9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.self_attn.v_proj/_s.pt
index 3208efdd5b50106b5501a2667f4309f1a1c55cbd..1796282d8784d870a48c6dd864caf036e0a91d63 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.8.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.fc1/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.fc1/_s.pt
index a6a937c14625ad60b2de0aa64a19d41a16f481da..36c77079f1039c23ec95246475fc016df29891d9 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.fc1/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.fc1/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.fc2/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.fc2/_s.pt
index 3505d4ff15d93bfa9cb3eb0803adc4e95e2c8485..5473f4c0d9e1147e06491a53a77295ae969b541c 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.fc2/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.fc2/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.self_attn.k_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.self_attn.k_proj/_s.pt
index 3b93782167ef136118a9471d1eca0e0f7a1fc2df..ee225377908ec0af2430b5f849846fe24f53bd24 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.self_attn.k_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.self_attn.k_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.self_attn.out_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.self_attn.out_proj/_s.pt
index cb6b13d8d7145e9e0a38572e333246d317fe3070..e41d3f74959b4dbf6d5d1897ce93bd09943971aa 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.self_attn.out_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.self_attn.out_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.self_attn.q_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.self_attn.q_proj/_s.pt
index 472d3b4020789648fc02933e7c9f6cffa91d8035..2e80edae61a359be6e81018543e749247fc5a6aa 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.self_attn.q_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.self_attn.q_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.self_attn.v_proj/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.self_attn.v_proj/_s.pt
index 130c36b63e826baa3c8d0a89d85279b24e6c6ab6..c04ed8f7a6e1fb8e86f1fcb537d11099b2391444 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.self_attn.v_proj/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.layers.9.self_attn.v_proj/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.project_in/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.project_in/_s.pt
index 430a9a06b9e35bc6cfbf24b3a42c6914b49da741..d39b2a02114c0c5306332b37a6a7314660b85198 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.project_in/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.project_in/_s.pt differ
diff --git a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.project_out/_s.pt b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.project_out/_s.pt
index cfca0e0c6b04d355ae70e9c3e1da74a1d387cca3..5a99160ecaf19baa579e15c224694f59d64e0c77 100644
Binary files a/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.project_out/_s.pt and b/params/opt/5/rnd_search_t_cos/comb_17/trained/model.decoder.project_out/_s.pt differ