diff --git a/flops-based/gpt2/1/0.pt b/flops-based/gpt2/1/0.pt
index db1c37b7a0d92873a74b25871f8db13bb77189d1..010e173fdfdbbe14c52218695499ef89e4bdfb41 100644
--- a/flops-based/gpt2/1/0.pt
+++ b/flops-based/gpt2/1/0.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:05db6a9a10931d0f57867380765a5674b57363e47153d1e20ccd8fecab581caf
-size 510726585
+oid sha256:8c304fd5feac770deddbdd9993b274cf21a5c9bc47a9fdea70df752a9d5124e3
+size 510726905
diff --git a/flops-based/gpt2/1/info.json b/flops-based/gpt2/1/info.json
index a6ffefa658caa5d3f6467643e6facda98f3d406d..b6cdc240d6151147e195fdb3d6a3c39f7f083112 100644
--- a/flops-based/gpt2/1/info.json
+++ b/flops-based/gpt2/1/info.json
@@ -14,7 +14,7 @@
         1607164416
     ],
     "times": [
-        0.03170204162597656
+        0.022766590118408203
     ],
     "params": [
         124439808
diff --git a/flops-based/gpt2/10/0.pt b/flops-based/gpt2/10/0.pt
index 9a34e305608d7b7d3dd77ab10ef0ce97dbcbc682..70176fd3e1f2bdf1c750950d6dc9c51961c3f770 100644
--- a/flops-based/gpt2/10/0.pt
+++ b/flops-based/gpt2/10/0.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:beaedef818d019813dd82f8fe24d2b36a153363bab45c8a16c8ffcf44ffceabe
-size 186978251
+oid sha256:ec0fd6b746c9ff21a8762f5e0530b96623e2d91deb24e53f42ac5b4f3bd2212c
+size 186978287
diff --git a/flops-based/gpt2/10/1.pt b/flops-based/gpt2/10/1.pt
index bc0acc6c15511aa81c4eaad4aa5fce92e2846d01..2b896205c0e4d40601c94f62cd8af6b9602795fe 100644
--- a/flops-based/gpt2/10/1.pt
+++ b/flops-based/gpt2/10/1.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:899eea1a6488c97416e50f14919d95ef37538c41c45b443bc9230d27449f9d17
-size 29434979
+oid sha256:78bf9c89d480ec74c3034d974d46409e642a893eb35855ba4f448a78fb275221
+size 29435067
diff --git a/flops-based/gpt2/10/2.pt b/flops-based/gpt2/10/2.pt
index 1d93370faebb83639dd32748e6e28157916622e1..872a3ca9077bf5703ea446eb8818cf11673d3d2b 100644
--- a/flops-based/gpt2/10/2.pt
+++ b/flops-based/gpt2/10/2.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9f160d87d7ee319dff5af72fa826987418b303beb89d830a4c7fba6dd4d045e3
-size 29434979
+oid sha256:592bedc00952f4a94a456bc5c9ab320130af298fd9aa043d396d418d7a51b972
+size 29435067
diff --git a/flops-based/gpt2/10/3.pt b/flops-based/gpt2/10/3.pt
index 027067841c88daf1f14dca1ad59e6e73dac014e1..03599f77a46518a8e2176124f3c7903f94efbab9 100644
--- a/flops-based/gpt2/10/3.pt
+++ b/flops-based/gpt2/10/3.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b510182b4a1e798b2cd4242d243bfffdfb083bad11b6e5ea737c1f3a1bdcf6e8
-size 29435043
+oid sha256:dab95015637342d81c946b92dcf69b76d0b1a80241bce016a22d10e9c19627c2
+size 29435131
diff --git a/flops-based/gpt2/10/4.pt b/flops-based/gpt2/10/4.pt
index 97055f62fdb494cd4aa9e8dbdd973df8747e961a..daa20f822d1d183365620f7db0ad740fd3daa483 100644
--- a/flops-based/gpt2/10/4.pt
+++ b/flops-based/gpt2/10/4.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3daf7ff37383dc2ffbf91f75c2ef37d46e36625361e6a3096328cf7717c6c7fe
-size 29434979
+oid sha256:5b5b138d60e0dcf1e47c0a0dc30464ca983b19e4298f1a7390f735914d54800d
+size 29435067
diff --git a/flops-based/gpt2/10/5.pt b/flops-based/gpt2/10/5.pt
index b74dc31b6818cdb64d61a758c1c70cee8b6d1895..b699b55844e685732c2aa7746c3b2aba1e76f6d4 100644
--- a/flops-based/gpt2/10/5.pt
+++ b/flops-based/gpt2/10/5.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:51f7d034831e92b70fe25d33f47925a9d60a68be2003af6c83e2fdfeae163bea
-size 58865457
+oid sha256:dadcae09fb03caad991674b65d0c815b73a76a8f454ffb815ae90800ade9ae5c
+size 58865557
diff --git a/flops-based/gpt2/10/6.pt b/flops-based/gpt2/10/6.pt
index c109c8d8eb1174ba6d51ceba7cd7f948894e63d4..c449d87b9a61ce12a2e08002bc6bfa1fd4eaddf4 100644
--- a/flops-based/gpt2/10/6.pt
+++ b/flops-based/gpt2/10/6.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:699287c8a1693920edb709b62d7be633f8c56781c97395af1ff9665bf1c699fe
-size 58865393
+oid sha256:29ee4bd1b483941a829fb1434f7ed747358bf46cf2a330a92eddc0d6497fdb03
+size 58865493
diff --git a/flops-based/gpt2/10/7.pt b/flops-based/gpt2/10/7.pt
index e2ecde1dc0fc6e71dc820b88f4235b806cc5d2aa..67f162ee4f04f9b42eb717f90e1c7503dd046212 100644
--- a/flops-based/gpt2/10/7.pt
+++ b/flops-based/gpt2/10/7.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2384c91641d076895d8ddeccd9e9b429b7292e3e04d1b71498cefe93f2716e8d
-size 58865393
+oid sha256:e194ec94e2046730f6b65b6ce707fe474d3d721ae87b8881401d1ac604a7371f
+size 58865493
diff --git a/flops-based/gpt2/10/8.pt b/flops-based/gpt2/10/8.pt
index ed4777ccb0309fb598eef241d53433a6b9953295..02d64b823b792d900d49755666424fcc0183b4a5 100644
--- a/flops-based/gpt2/10/8.pt
+++ b/flops-based/gpt2/10/8.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:63572711e7faa8c364e36bc7de8cebb6f100a18ba0035a06f939df306b0df77b
-size 29444359
+oid sha256:f80c3143fa7e74a0702d806c470df228290793671342ea4d67cbed3ffe50bda5
+size 29444523
diff --git a/flops-based/gpt2/10/9.pt b/flops-based/gpt2/10/9.pt
index e14936e40778c0aa6dff0d6e2fff60e647eec54c..bf8988145f5e5c7e1d36df747b0628aeb3cdb62e 100644
--- a/flops-based/gpt2/10/9.pt
+++ b/flops-based/gpt2/10/9.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7628aba996bdaa95e0ec16179361b30830ee026665a896c3a4d3ccd62c74de78
-size 154395921
+oid sha256:d3df76c2b6db56fbca7ffe8a60992cb791c22b203eaa28ca507c2aaf253bd772
+size 154395945
diff --git a/flops-based/gpt2/10/info.json b/flops-based/gpt2/10/info.json
index 3fa8428d8aff9b2d9e5b0a84aec7ebe6c71bd314..4cab6f975b0a9bc75197c4822de7e84c400736f6 100644
--- a/flops-based/gpt2/10/info.json
+++ b/flops-based/gpt2/10/info.json
@@ -68,16 +68,16 @@
         501765888
     ],
     "times": [
-        0.0019626617431640625,
-        0.002095937728881836,
-        0.002147197723388672,
-        0.0018086433410644531,
-        0.0051004886627197266,
-        0.004480123519897461,
-        0.003067493438720703,
-        0.0031020641326904297,
-        0.0015683174133300781,
-        0.004399299621582031
+        0.0023927688598632812,
+        0.0019953250885009766,
+        0.0022687911987304688,
+        0.00250244140625,
+        0.002256155014038086,
+        0.0032486915588378906,
+        0.004351377487182617,
+        0.003931522369384766,
+        0.002094745635986328,
+        0.005228519439697266
     ],
     "params": [
         46471680,
diff --git a/flops-based/gpt2/11/0.pt b/flops-based/gpt2/11/0.pt
index 37d2108cf2c3da7a83784287d56cb1c0d8336bd4..8a5ad00b9929b9278c620cbbf697d3dfe04faadc 100644
--- a/flops-based/gpt2/11/0.pt
+++ b/flops-based/gpt2/11/0.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d0e0c740d28f4714b4c83a4081c1d9f1f8be14ed437c675b9d91d7ba35eda001
-size 186978251
+oid sha256:87fd1f2326ea0aff3d4fd5b5585c9e0ffbe10ad9d77a6ba43bdfc9767c091d0f
+size 186978287
diff --git a/flops-based/gpt2/11/1.pt b/flops-based/gpt2/11/1.pt
index f7114233a92ea290a64bcb732e955ddbc93a0a3c..d1239dac5bb4b0aa48c6e72c860ef09d36fd5cf2 100644
--- a/flops-based/gpt2/11/1.pt
+++ b/flops-based/gpt2/11/1.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d29b0a93190c6fb1fb77c44e4a2017d48a99b2ca4b5ccacd7a38848d4d1783c3
-size 29434979
+oid sha256:5d56fc7b5a26673ac514ab72e8712692847f314b9cb0b7e30ad0c1c9f8e4668d
+size 29435067
diff --git a/flops-based/gpt2/11/10.pt b/flops-based/gpt2/11/10.pt
index fe24c0406ef2bf4d389788b290d401ddc52d8aa9..17df5393e6c6a7538e5d7004c11347aca330ad87 100644
--- a/flops-based/gpt2/11/10.pt
+++ b/flops-based/gpt2/11/10.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3e22fd2000d55c7398d90726ab10a5e6e3f0878ddf349ec62816e9952507b6d9
-size 154395937
+oid sha256:e478302f030cef5255cfcdcec188ff148167b8caebec7cf80dc0d2a951322d6a
+size 154395961
diff --git a/flops-based/gpt2/11/2.pt b/flops-based/gpt2/11/2.pt
index 144adeb929162e243c91eeb32f8fb98837d19650..bb14e9aeda1e04ed6c501664f516e11dca42ce0b 100644
--- a/flops-based/gpt2/11/2.pt
+++ b/flops-based/gpt2/11/2.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:863656105749d4aa0abcb352630b6ff58df3f6b592c744ffdf7944f8366a4878
-size 29434979
+oid sha256:87345183eae9665ee48f676c2c06c8f0502ab5cdea65aea167d0f586c3f134ea
+size 29435067
diff --git a/flops-based/gpt2/11/3.pt b/flops-based/gpt2/11/3.pt
index 9de1c9b65174b5e8bb949047b48c314d82a6beec..b4bde2324262af45620da515eb9d1c81e5d05f1a 100644
--- a/flops-based/gpt2/11/3.pt
+++ b/flops-based/gpt2/11/3.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0077af0b70966277abd01a76c9c31aa27b310526a82ed252b4f0bdbcf803741a
-size 29434979
+oid sha256:8febf8c44176801783d647af69b877a36a4a8511999d06f93d2a790b6c48d2e4
+size 29435067
diff --git a/flops-based/gpt2/11/4.pt b/flops-based/gpt2/11/4.pt
index 701c63775986c9ab1cd1b9b1878e3ccf94e08fa1..df308bcbe744e6f44d270d7d08ea3e02768a4cf0 100644
--- a/flops-based/gpt2/11/4.pt
+++ b/flops-based/gpt2/11/4.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e0021fd79c2e43fed69dceeb3c608f07041c475e774ae367f2bf1e718b6866d9
-size 29434979
+oid sha256:a47a6aca3c0feab5e3b2306d0c00e31405e94d0e7fafd6fd2c9aef430d5e0fd5
+size 29435067
diff --git a/flops-based/gpt2/11/5.pt b/flops-based/gpt2/11/5.pt
index a6347ad64c5517e48b1e3d8c0dbebc37fa085347..1cb48c630b22c174f76841488662028ea047e786 100644
--- a/flops-based/gpt2/11/5.pt
+++ b/flops-based/gpt2/11/5.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8be8709f576d042ead24f9415b570642101197920152adcc6815c7a85177abba
-size 29434915
+oid sha256:6b827b9d4e1f1da385c1a89fb9d4d0abb09a1814f40f22ab64addaef838bb04b
+size 29435067
diff --git a/flops-based/gpt2/11/6.pt b/flops-based/gpt2/11/6.pt
index 3d58226d383b3c4a506e6c09ce4cbcd84eba11a6..8c413aea347e8827e826e0e27cb6c15d81b4b63b 100644
--- a/flops-based/gpt2/11/6.pt
+++ b/flops-based/gpt2/11/6.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7eef8a1c9d81b76db024a8136cdfacd7141bcb76d89af772959af0ee5493af77
-size 29434979
+oid sha256:c7aa3f815f9f6d1a20a76248a113dcbddb439a128c7dc6210e12d25897616039
+size 29435067
diff --git a/flops-based/gpt2/11/7.pt b/flops-based/gpt2/11/7.pt
index 06f2a9ffe5b332beed05ddbaf1673f6363c58902..c01390baf717dac88bcdf7c58ff3e940dff686bf 100644
--- a/flops-based/gpt2/11/7.pt
+++ b/flops-based/gpt2/11/7.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4d4ea660b9928a4903985ba23569de60334e0b432edf110e339d1e4cfc9dc1df
-size 58865393
+oid sha256:679ae435b52b258232864f15b558c66210dd968fd017097164a76822af30543e
+size 58865493
diff --git a/flops-based/gpt2/11/8.pt b/flops-based/gpt2/11/8.pt
index 8fe99930b8bfd6d597d28eb36c566731921e35ad..2eae727e5e461bcc12eebebb93493005d7eb62d9 100644
--- a/flops-based/gpt2/11/8.pt
+++ b/flops-based/gpt2/11/8.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:324709eab430937a433276dfafefb7d298ab118c6f99833e85629131b5ff1866
-size 58865393
+oid sha256:cbead182734f5020d831f633df884aeea695e60c38f8a0782ef3feacd21f52ed
+size 58865493
diff --git a/flops-based/gpt2/11/9.pt b/flops-based/gpt2/11/9.pt
index b9964a70279a782c08a6d24b6b645bd054dc05f4..f108b0af950018bc8079b67061ef50373acb1cb2 100644
--- a/flops-based/gpt2/11/9.pt
+++ b/flops-based/gpt2/11/9.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:45909f3cbf63417d14d3cc7777b9fe3864e4bf8c2e060a10a8720bf4fd281d40
-size 29444359
+oid sha256:0e0ba1fab6a2edf8625f1538452693a359112dfb3bc00059392792011bf5d99b
+size 29444523
diff --git a/flops-based/gpt2/11/info.json b/flops-based/gpt2/11/info.json
index a8a65f591f23d0cd18a2e4ecb73c24a0b1957291..27675c8684bf0f0c4158d725e816cbc413320df1 100644
--- a/flops-based/gpt2/11/info.json
+++ b/flops-based/gpt2/11/info.json
@@ -74,17 +74,17 @@
         501765888
     ],
     "times": [
-        0.004611968994140625,
-        0.0019762516021728516,
-        0.004351377487182617,
-        0.0023436546325683594,
-        0.0022706985473632812,
-        0.0025353431701660156,
-        0.0018358230590820312,
-        0.0043811798095703125,
-        0.0058441162109375,
-        0.0034186840057373047,
-        0.004660606384277344
+        0.0024166107177734375,
+        0.003831624984741211,
+        0.003988742828369141,
+        0.00365447998046875,
+        0.0035195350646972656,
+        0.002308368682861328,
+        0.003462553024291992,
+        0.0036993026733398438,
+        0.0059032440185546875,
+        0.0023107528686523438,
+        0.004665374755859375
     ],
     "params": [
         46471680,
diff --git a/flops-based/gpt2/12/0.pt b/flops-based/gpt2/12/0.pt
index bc6c54e776d3831f7f3720a9107a3921bdaf64d1..5293ad83bd03e8c4c490e2e7055d918915ef6ce1 100644
--- a/flops-based/gpt2/12/0.pt
+++ b/flops-based/gpt2/12/0.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cff068d37f141ea34d5663038ce86e6fb532f695ea74ef7dbc4ec1c0ea0893cc
-size 186978251
+oid sha256:9f6bb3fc75e04e125f3fa119614ad1f53bdd62073a69386b3c86ef602de722fb
+size 186978287
diff --git a/flops-based/gpt2/12/1.pt b/flops-based/gpt2/12/1.pt
index 97003b5404ef2abd81ae2107273f3cae98770806..46f9a85f4c6fdd2e4aebc3c433ed553e998d1711 100644
--- a/flops-based/gpt2/12/1.pt
+++ b/flops-based/gpt2/12/1.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8e14b009300745fdda3be9cb0165e73f4a36182388e3298b76c80a7abd8e548c
-size 29434979
+oid sha256:86da82964f62e1ad26d021286450756788e05f09ed58fbadca1b46a3dffc8ce7
+size 29435067
diff --git a/flops-based/gpt2/12/10.pt b/flops-based/gpt2/12/10.pt
index 23b34ef4e684d41d8cfe2b84b45f11e7c7b38ef2..5987071c843370d8b0825058ec0537dbdb2dbce9 100644
--- a/flops-based/gpt2/12/10.pt
+++ b/flops-based/gpt2/12/10.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:46e133126a55b8d79974c0c4e034d489e5ba364b9e46045c7c78b544bb752f27
-size 29444616
+oid sha256:c42b911603bb3f21a8d949e48e1b4f9e718f8730b34448cec9d903cf0593b9b6
+size 29444716
diff --git a/flops-based/gpt2/12/11.pt b/flops-based/gpt2/12/11.pt
index 60174895d2a8e63c4c745de7f5ca7cf5ed1086a0..a2333058be2a1aa6f7403cf76c857eae1c1867c4 100644
--- a/flops-based/gpt2/12/11.pt
+++ b/flops-based/gpt2/12/11.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dc288f96064f96ccecf4ebd5de566797b74fff0cf61770242829c71f7ab8e81c
-size 154395937
+oid sha256:4a090e75c31929ab85bbdc9ffdc2e8f5d397266fe596659b46d972c5274a743c
+size 154395961
diff --git a/flops-based/gpt2/12/2.pt b/flops-based/gpt2/12/2.pt
index b794a7b40e81fd2e7a97adda604eefe4525b69a8..62ce58c01e14124dbe737103ae9307a41b2dc03b 100644
--- a/flops-based/gpt2/12/2.pt
+++ b/flops-based/gpt2/12/2.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a140235023877874a04e673c0031519caea48411ad85b742bf0fa479cda18b83
-size 29434979
+oid sha256:0d49d76a0264bf7dadd8abb025e48087ad60e8ed513adca56eac457aa0ceb74e
+size 29435067
diff --git a/flops-based/gpt2/12/3.pt b/flops-based/gpt2/12/3.pt
index bdc91dd14c24b9bd815f4d5dc218f12205b949b9..1ec6ec30e93696a14781c8b3cd3262ce71d92321 100644
--- a/flops-based/gpt2/12/3.pt
+++ b/flops-based/gpt2/12/3.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9c9f119807048c2bbd1e92f1ddf9c6f3661b18bee2e55d68d799a86eba160c3d
-size 29434979
+oid sha256:90b2a73767b9ebcd8718e57cafb4c1c09287d3b0e122b6870b11fd0240985d63
+size 29435067
diff --git a/flops-based/gpt2/12/4.pt b/flops-based/gpt2/12/4.pt
index a0372c4133d696dea90d327bb533e90df602475b..41223e6551704e1865a4c4d4c16b5220f9c84098 100644
--- a/flops-based/gpt2/12/4.pt
+++ b/flops-based/gpt2/12/4.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2642a80a09607c273532bf0a8f005693392a00ec99e4f0dec1549e5bc5afc582
-size 29434979
+oid sha256:85697de35ba8e100e8712fb851e24fa3ad772e6510dac009360bc27d747d6012
+size 29435067
diff --git a/flops-based/gpt2/12/5.pt b/flops-based/gpt2/12/5.pt
index ec5c79eff837d442201a3c558264b7b219a89cb6..fc137ca44856044a904cf5f89a0ba96c06defcc4 100644
--- a/flops-based/gpt2/12/5.pt
+++ b/flops-based/gpt2/12/5.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:521fb2a199fd882395862aabada7a98cddfbbbfba4550b0f751dbb5e96d1e177
-size 29434979
+oid sha256:559948526b479f93575d248b991724c4c8249e7cb88ddec9aa20aff7626fecd0
+size 29435067
diff --git a/flops-based/gpt2/12/6.pt b/flops-based/gpt2/12/6.pt
index 0f7e99a0124a91b3d85c8da5f3d8f322432feab5..51b1a63b8882996f708e486ea3722f481c663682 100644
--- a/flops-based/gpt2/12/6.pt
+++ b/flops-based/gpt2/12/6.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e4ed21d0f18aad2987d1d1cfdef442e725006336cd7632c1a072b6498e81c746
-size 29434979
+oid sha256:36fec45fd089b3660ed0a308f021429ed4be83889af31c93fd78c0375e457211
+size 29435067
diff --git a/flops-based/gpt2/12/7.pt b/flops-based/gpt2/12/7.pt
index 7336514bacde1bf7c3d4a511dc13ae03a85bcc06..1bcb6edda0e119dcc406eaf942c5fd579514c0f6 100644
--- a/flops-based/gpt2/12/7.pt
+++ b/flops-based/gpt2/12/7.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:539ae899cb1c02f7f42019b6d68a4f1875c9f6461ea431e3a22078bb71bef107
-size 29434979
+oid sha256:1576718ccf4be4aa4cfa275ca28bfc84d4c2bcc417aa58b59550bdaf146e5dfd
+size 29435067
diff --git a/flops-based/gpt2/12/8.pt b/flops-based/gpt2/12/8.pt
index fc5e017ca4c25d57c5edc6ff44a066b62597067e..09f1e38289c4236f962ee49c23216b5cde86fc11 100644
--- a/flops-based/gpt2/12/8.pt
+++ b/flops-based/gpt2/12/8.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c8aff174abd91da203995f6a1e89cc004b5a1d675be0a22abb6bcbc89865db6a
-size 29434979
+oid sha256:d09e4b1f452d3716f2517915d299787e8f89e75bbff93d7e10ffd1b24fffaad7
+size 29435067
diff --git a/flops-based/gpt2/12/9.pt b/flops-based/gpt2/12/9.pt
index 240022cb2f8089403b17f620b19d7893ab0bb9d3..342676a082b3ea2c0fea7a1c1a824a8e3efd7f1a 100644
--- a/flops-based/gpt2/12/9.pt
+++ b/flops-based/gpt2/12/9.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0550f5ba97d3b337c5067084886ca7f2772c2f26adfb8d18ab513dd3fe93a423
-size 58865393
+oid sha256:9bfc4fe5421abc9a29b8cde16e356547adf1f6bb5f0141333c8302549a230f6e
+size 58865493
diff --git a/flops-based/gpt2/12/info.json b/flops-based/gpt2/12/info.json
index 5b86dd15f76deedd11bdb5d43452ba9ea7a31f6f..51e0735bc50bc98510d8f5f9ec4f80cbdc3384c8 100644
--- a/flops-based/gpt2/12/info.json
+++ b/flops-based/gpt2/12/info.json
@@ -80,18 +80,18 @@
         501765888
     ],
     "times": [
-        0.002033233642578125,
-        0.002433300018310547,
-        0.002496480941772461,
-        0.0023779869079589844,
-        0.0028944015502929688,
-        0.002802133560180664,
-        0.0023040771484375,
-        0.003048419952392578,
-        0.0022711753845214844,
-        0.003820657730102539,
-        0.002216339111328125,
-        0.0046138763427734375
+        0.002263307571411133,
+        0.008869171142578125,
+        0.002043485641479492,
+        0.0037772655487060547,
+        0.0022172927856445312,
+        0.004255533218383789,
+        0.002051830291748047,
+        0.0020220279693603516,
+        0.0025339126586914062,
+        0.0047757625579833984,
+        0.002246379852294922,
+        0.004633665084838867
     ],
     "params": [
         46471680,
diff --git a/flops-based/gpt2/13/0.pt b/flops-based/gpt2/13/0.pt
index adf78d317b78792bdc22334bfea8ed2afdd50190..37665e361729ca31a5c3a0944c228cfa85a0e27d 100644
--- a/flops-based/gpt2/13/0.pt
+++ b/flops-based/gpt2/13/0.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:df5c8b0f7e7617e5881e8af29ee251db74be90c80fb8d4962d48f69976f06750
-size 186978251
+oid sha256:224d313091361a98b92b181b91fcda9dd12dc35407d4cefaeda7d9a0d4ec3b30
+size 186978287
diff --git a/flops-based/gpt2/13/1.pt b/flops-based/gpt2/13/1.pt
index f683efb348485c84c7d258839495f1a6c0025a57..3aab150cf04826478608c36e7cdc87e39b1fbdcd 100644
--- a/flops-based/gpt2/13/1.pt
+++ b/flops-based/gpt2/13/1.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:71f3dfe539a81d38f0b3a006a40152c35eaf80a3821d5ea2d30b43edf1b1910d
-size 29434979
+oid sha256:b81e83da135dc8b0becfc45761fa7ebb76aea11cdb8030132b35e99fdf93067b
+size 29435067
diff --git a/flops-based/gpt2/13/10.pt b/flops-based/gpt2/13/10.pt
index ac2e75443ad58358289669a3a4e2ef179645344c..593fde3660d1f06facacb1afb70581f65100916e 100644
--- a/flops-based/gpt2/13/10.pt
+++ b/flops-based/gpt2/13/10.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a1b20980b7787e04130d4a855e986c0d41495d1df16303459d17ac56e1e1eb6e
-size 29435166
+oid sha256:4ece51cef72c34ef85dd2ec0877baac0f353927d60ba6bbe31653cb69ca39d19
+size 29435254
diff --git a/flops-based/gpt2/13/11.pt b/flops-based/gpt2/13/11.pt
index d85c28cfce56541b764e2a269804d8498b0861fa..8a134825d6f3981714a7f350697094e4a7e6c43a 100644
--- a/flops-based/gpt2/13/11.pt
+++ b/flops-based/gpt2/13/11.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bed054cf5b775e61b4fd2349bf7d1e73f271516450a543bbdb8c6ebb0508cd96
-size 29444552
+oid sha256:b9674a2601e65c3b0715329245c710131951741129cd1e33b250b0966e38fe66
+size 29444652
diff --git a/flops-based/gpt2/13/12.pt b/flops-based/gpt2/13/12.pt
index 63ecfe2dd63c9537ce41816cb20bf9d4a66fcd7e..9dc25c948a04b50df8af72636a1755672651bec4 100644
--- a/flops-based/gpt2/13/12.pt
+++ b/flops-based/gpt2/13/12.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a0b85dfcec025ec3d6bb2cb89ef46a3f8fd374fcce19d0b635ab71b1485823eb
-size 154395937
+oid sha256:a6b0c5c1ac94c0845c458b8be9273534531be7274eb9a0d3c89a6c3cd2104584
+size 154395961
diff --git a/flops-based/gpt2/13/2.pt b/flops-based/gpt2/13/2.pt
index 34fec8d82ffb44d3704cf957b189a95ee43423b5..96a9aae1aadc033a192c84e5438a816268c7d7b6 100644
--- a/flops-based/gpt2/13/2.pt
+++ b/flops-based/gpt2/13/2.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:be01308db019db3cb48cb47fe6cf4cbbc8769fa38045ea7161ed28ee3b1392da
-size 29434979
+oid sha256:9355b51b332c1dc18c6bd8f4968397fe173a0b001d04f1f44db4d6de67d17fad
+size 29435067
diff --git a/flops-based/gpt2/13/3.pt b/flops-based/gpt2/13/3.pt
index 85f633242d01b0b5e677f1c4786ef84c48a29ae7..d7b4e62017707f2f2241bfd4044b90ac60998772 100644
--- a/flops-based/gpt2/13/3.pt
+++ b/flops-based/gpt2/13/3.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9508bb20a32cc652808b52aec7e72977755a606cd472e00a2845a51246f892b3
-size 29434979
+oid sha256:2dbe21ebc29fa01c783c935e9ed61f2ec0a74e5f79827226a562689added356b
+size 29435067
diff --git a/flops-based/gpt2/13/4.pt b/flops-based/gpt2/13/4.pt
index 37b86b7a961a4e69d96c803ff2e1cf906655f740..c73e760202ee5ffdde6daa981620fbda446a4710 100644
--- a/flops-based/gpt2/13/4.pt
+++ b/flops-based/gpt2/13/4.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:05bc86295e6325dac725f63338b9e9d8eb07f53d3143117a1b8663bc25d91bf4
-size 29434979
+oid sha256:06f2a998015ee35f53344c68af74457d2e336c9e0f3797548ee770ab27ff3eff
+size 29435067
diff --git a/flops-based/gpt2/13/5.pt b/flops-based/gpt2/13/5.pt
index 1475a674fc8f91c3cb49c989a1348c48adc9982a..705dd9720c524e335642acdbdb592c7301955b6a 100644
--- a/flops-based/gpt2/13/5.pt
+++ b/flops-based/gpt2/13/5.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c3ea5ce1c92d208f1d5379eff94823f7827c7a6c8316934536562e9173c9dc93
-size 29434979
+oid sha256:c255cb2f11602524b695cf10b2b86d48a8a2527928deee874da6cea8d3bdd47a
+size 29435067
diff --git a/flops-based/gpt2/13/6.pt b/flops-based/gpt2/13/6.pt
index 65dacc60373a3bf81ea460e04ae93a44d15228ad..3e23af637abb7b51cb2be57c1545f01159585cf3 100644
--- a/flops-based/gpt2/13/6.pt
+++ b/flops-based/gpt2/13/6.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bd430343a81fb9a328910c4eabbfb98a464c6ec958460ecee0e295c1d6ab5abb
-size 29434979
+oid sha256:c15252f3930ad4832545742db8e209f9c684cdd50c9ce738e7a4167a42290a12
+size 29435067
diff --git a/flops-based/gpt2/13/7.pt b/flops-based/gpt2/13/7.pt
index b260ed3e61a0b71605ce4bc0e15ae22440668848..965139a24d4879c898863653203019cb12171da1 100644
--- a/flops-based/gpt2/13/7.pt
+++ b/flops-based/gpt2/13/7.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d606e4ebced1dcb14a8fcf67abb1cb8088da47b36c8d3f6d7085ca3c34125dcf
-size 29434979
+oid sha256:e59e13e4cab18bb564884b5620ac333806f9045108104a534fb1f3f4098a2372
+size 29435067
diff --git a/flops-based/gpt2/13/8.pt b/flops-based/gpt2/13/8.pt
index 5047b904b4aed0607fd7c72220402c8dd9cd754e..7e7941ca7872aa1aab78abd198de940a92711dec 100644
--- a/flops-based/gpt2/13/8.pt
+++ b/flops-based/gpt2/13/8.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f83d6ab04b5887f7683244b1ca8f8096dc897ff8d9d6a10773285f08f34ecb95
-size 29434979
+oid sha256:9c354337ebb059f3957e80de86407bee34764ea07887201cc6df7d5c4b81e0fe
+size 29435067
diff --git a/flops-based/gpt2/13/9.pt b/flops-based/gpt2/13/9.pt
index ed99e7ef1029d16e40240fe0e7d2e69eb78ec3f1..9da7ea95a5059d4621372776fb2960d2a12582e1 100644
--- a/flops-based/gpt2/13/9.pt
+++ b/flops-based/gpt2/13/9.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aad4a85f5fa87083d67b8bae7fd1637e0644cec43e43d8fe6383fde54f3d6952
-size 29434979
+oid sha256:8d35d56a58abae15f6aedaf06086ba9cfb4fe5d7829e0c03cf95722f8ae36f89
+size 29435067
diff --git a/flops-based/gpt2/13/info.json b/flops-based/gpt2/13/info.json
index 7068d5be441a8e01da4a28206de4fe7cada72733..81b0bd0fa6c8dd60aa162da7cf48220c76aaf5a7 100644
--- a/flops-based/gpt2/13/info.json
+++ b/flops-based/gpt2/13/info.json
@@ -86,19 +86,19 @@
         501765888
     ],
     "times": [
-        0.003165721893310547,
-        0.0020742416381835938,
-        0.002318143844604492,
-        0.004423379898071289,
-        0.0019943714141845703,
-        0.003233194351196289,
-        0.002018451690673828,
-        0.0021936893463134766,
-        0.0034558773040771484,
-        0.005201101303100586,
-        0.0018916130065917969,
-        0.0032739639282226562,
-        0.0044727325439453125
+        0.003621816635131836,
+        0.006613492965698242,
+        0.0036499500274658203,
+        0.0036177635192871094,
+        0.003648042678833008,
+        0.011776924133300781,
+        0.004663944244384766,
+        0.003574371337890625,
+        0.003004789352416992,
+        0.0020940303802490234,
+        0.005176067352294922,
+        0.0023741722106933594,
+        0.004660367965698242
     ],
     "params": [
         46471680,
diff --git a/flops-based/gpt2/14/0.pt b/flops-based/gpt2/14/0.pt
index ac9ea96dd7e3a183b36a7447ee95c4b022ff8d28..ca2e85fb8dab81c6c710e8d77adbc1a3595f71dc 100644
--- a/flops-based/gpt2/14/0.pt
+++ b/flops-based/gpt2/14/0.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5093289c7109074ec3f726f2cb604593f630dd3dc1a66043e32c873303963add
-size 186978251
+oid sha256:043e2cf4de2b5a51303a3181fe131a83a4a7271eaa3e16b85858bc2e47338fe5
+size 186978287
diff --git a/flops-based/gpt2/14/1.pt b/flops-based/gpt2/14/1.pt
index 84f57c6e52d21848432d544ec177eb0fe27fccce..21ed3e73d032e06c3a6bfc8dc2507ee93c10509d 100644
--- a/flops-based/gpt2/14/1.pt
+++ b/flops-based/gpt2/14/1.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1bfbc5ec45da1840b9a0fe26d328576df164983ab4fe03ebdef6eaa041bd7a3f
-size 29434979
+oid sha256:2574f4a149ee45730de4a3e4feec5004891699d3a4bcbfaa41698b50d5d92015
+size 29435067
diff --git a/flops-based/gpt2/14/10.pt b/flops-based/gpt2/14/10.pt
index ec9e7c024197cd1283ddf1ab27c4451d59ae9d19..9b6a33560092ea2b3664ca7ddf4bd9c22b882f4b 100644
--- a/flops-based/gpt2/14/10.pt
+++ b/flops-based/gpt2/14/10.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:439b9266eb2c5dad6fcb326dd4e060a273156bdd65f349f73286829ed3203eee
-size 29435166
+oid sha256:43828ae5893f72d2106af90e03559361f2b37fbd3c4caf2249de6e4f71760aa3
+size 29435254
diff --git a/flops-based/gpt2/14/11.pt b/flops-based/gpt2/14/11.pt
index 7c68ddf246543d156b9449574a20a0bd457fceea..d12aa48a9682899bea622ab1e82312c080f4945b 100644
--- a/flops-based/gpt2/14/11.pt
+++ b/flops-based/gpt2/14/11.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:90113445773154694810f77fb0f9adff721ccdf330d3dc3b03e36221c2e5d312
-size 29435102
+oid sha256:a25cb5a6058c822d14cdd0a9643801530a409508a9b7cf9062e4663d08fce901
+size 29435190
diff --git a/flops-based/gpt2/14/12.pt b/flops-based/gpt2/14/12.pt
index eabbe0650c17fa1a603cc3b1e8b801b9e073242f..439d29ea6d12e23869498490c93df2700bc17c3b 100644
--- a/flops-based/gpt2/14/12.pt
+++ b/flops-based/gpt2/14/12.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7fd6d5f402c90a79a29e0394114b83ff2f8a9c8e0a05e4aa49259ed40699eb71
-size 12710
+oid sha256:32bfd7809e44cd821af84faa25c9f0d361e27b6efc74b75c051ee39789240597
+size 12862
diff --git a/flops-based/gpt2/14/13.pt b/flops-based/gpt2/14/13.pt
index 727947b794a1d8299065065aa8575ab7a240931c..48649f592e7da842db4222ebbfbb4f9e27bfe978 100644
--- a/flops-based/gpt2/14/13.pt
+++ b/flops-based/gpt2/14/13.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:abdd5d89b2544a60057b6f5feb29180ef3aa33502b43720ac7d1b952db39fc78
-size 154395937
+oid sha256:17c609defec2f68eb4193d8aa5b887972069a1a05de9093d77218b76798d7bdd
+size 154395961
diff --git a/flops-based/gpt2/14/2.pt b/flops-based/gpt2/14/2.pt
index 315684c97c2374e7a1d5ae44bec831a960b39185..e7646b36199ab170c84d9e72fe9b86cfdd5dabc3 100644
--- a/flops-based/gpt2/14/2.pt
+++ b/flops-based/gpt2/14/2.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6c2a286aaa934c79da45dc66d2c54d47dea00f15c8c293a0fbb8698fcba20718
-size 29434979
+oid sha256:dbf9a759ed80b4d663411322699c6aa97637497c9d0d3c0fe239309e10ad29c3
+size 29435067
diff --git a/flops-based/gpt2/14/3.pt b/flops-based/gpt2/14/3.pt
index 63d42456282cdda817eb22dbe563207fa13caeb5..09f8dd489f3e708e51f7f6d99d729ae53588c541 100644
--- a/flops-based/gpt2/14/3.pt
+++ b/flops-based/gpt2/14/3.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dde49c051f64384b494bc2ecad70f02137fb8a51413683398a14e9c5b35e04eb
-size 29434979
+oid sha256:0cc687451dd6a98659f7f6f7bf81abe7967de5910a813c4d2c48f6afb6986837
+size 29435067
diff --git a/flops-based/gpt2/14/4.pt b/flops-based/gpt2/14/4.pt
index bcd78a181265b01ad02363c69eb4de0433eea0ce..c8968306f06e5e8c92f6c05a8beb011864e52beb 100644
--- a/flops-based/gpt2/14/4.pt
+++ b/flops-based/gpt2/14/4.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a49bd3e495cb198811838a57ff3174de9eac7ca33b2dc33fdec2ea789d371630
-size 29434979
+oid sha256:2ef796a625c1f957a35f1a5916e14beced5b4877dd4c6d0178b9c193a9d649a7
+size 29435067
diff --git a/flops-based/gpt2/14/5.pt b/flops-based/gpt2/14/5.pt
index 5f9ab0a438b0e7182a52650b05af732f8a5924ab..1f5d1f4bdfeceae67831c01b0211fb804c33459e 100644
--- a/flops-based/gpt2/14/5.pt
+++ b/flops-based/gpt2/14/5.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:341cd6b7523e07e954f28a189a6f8f8d191cabd43ca750e6df397f4149bfeb01
-size 29434979
+oid sha256:8e9c38d3f531e6cffe03aded060580a35b0a25edc02c8a0d51f35273d91c143f
+size 29435067
diff --git a/flops-based/gpt2/14/6.pt b/flops-based/gpt2/14/6.pt
index 8f92fc23d5fe5e04710cb1fb04c32cd61dfb5129..5df4388873a1043d319dda3e388a7adb4512146f 100644
--- a/flops-based/gpt2/14/6.pt
+++ b/flops-based/gpt2/14/6.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f4d5261ade7ab7f3315fce728842a298f0ed8ee08e66c9a55658c35c97cb8c82
-size 29434979
+oid sha256:2792b7f5b4b174022ea544a2c7ce1a56363ee8dd1df606a9a0eaebd6f8cb383f
+size 29435067
diff --git a/flops-based/gpt2/14/7.pt b/flops-based/gpt2/14/7.pt
index 6c1b0e02949c0731f72dbf27e950c61d1d590dcd..17a65d042ae483673209a5505d5dac922cde8c78 100644
--- a/flops-based/gpt2/14/7.pt
+++ b/flops-based/gpt2/14/7.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f5aa964faccc0838a3b7aec170ad6d553f42cffdb1a02c275fafd241cc61d8cb
-size 29434979
+oid sha256:7bba26771cd7fc604380bba9a8860c4503c868c1df984c76654b608930c57a54
+size 29435067
diff --git a/flops-based/gpt2/14/8.pt b/flops-based/gpt2/14/8.pt
index 9e6e9553b0d811965ab62825108a2c09b1c3c903..26ca707f398899650448e65665fd8e89376e6e33 100644
--- a/flops-based/gpt2/14/8.pt
+++ b/flops-based/gpt2/14/8.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d51cecf662bc6ccee0deac862f11afb8ff4afcea8ed439a5b68713d303ee7f07
-size 29434979
+oid sha256:0d2707b332eba824f423c9ba962af33d4af5705cfce5970dfc309630a6a990e6
+size 29435067
diff --git a/flops-based/gpt2/14/9.pt b/flops-based/gpt2/14/9.pt
index 6d0ebcf689d251c4af9ceb25938ab0a2c2caacb9..6771e301534aa372b469342e77d7d4d47114816c 100644
--- a/flops-based/gpt2/14/9.pt
+++ b/flops-based/gpt2/14/9.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4e14daa9d4151f79e3133e7bcc5cc6dcdfe65db9769062c48431b6465fec180c
-size 29434979
+oid sha256:838d731be41e36d6ce344996a55023fc1abc4d4dc934cb0ae5aecb3af2feee41
+size 29435067
diff --git a/flops-based/gpt2/14/info.json b/flops-based/gpt2/14/info.json
index 7660394d3c89d06722552692857e47c22288a9ac..8cbd56a8330f6dff48cd69c39f00b3a1a1298ad0 100644
--- a/flops-based/gpt2/14/info.json
+++ b/flops-based/gpt2/14/info.json
@@ -92,20 +92,20 @@
         501765888
     ],
     "times": [
-        0.002893209457397461,
-        0.0025644302368164062,
-        0.002410888671875,
-        0.003850221633911133,
-        0.002217531204223633,
-        0.0028934478759765625,
-        0.005591392517089844,
-        0.0019474029541015625,
-        0.0022170543670654297,
-        0.0021944046020507812,
-        0.002183675765991211,
-        0.0045969486236572266,
-        5.9604644775390625e-05,
-        0.00461578369140625
+        0.0018732547760009766,
+        0.0018281936645507812,
+        0.002164602279663086,
+        0.0020360946655273438,
+        0.004652738571166992,
+        0.004725456237792969,
+        0.002803325653076172,
+        0.0020804405212402344,
+        0.0021009445190429688,
+        0.004415750503540039,
+        0.0034682750701904297,
+        0.004538774490356445,
+        5.8650970458984375e-05,
+        0.004797935485839844
     ],
     "params": [
         46471680,
diff --git a/flops-based/gpt2/2/0.pt b/flops-based/gpt2/2/0.pt
index f8772263ecff387b3c784c5c5ce8307871500244..48bb3324c96ba6a435f72acd21a4a95792aa3416 100644
--- a/flops-based/gpt2/2/0.pt
+++ b/flops-based/gpt2/2/0.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:db2f18303ff6ae29f3966913064200e796c7f90fd6e2e5d505eec85565432814
-size 422422551
+oid sha256:cfcc5c9750d6808ec396d934e483603055ce291141248477313bbcab9c217da7
+size 422422747
diff --git a/flops-based/gpt2/2/1.pt b/flops-based/gpt2/2/1.pt
index 1edbec9aae124ade36a3e5f92848ceb8d6272bc4..30b6d2c1b67a3f28439cb7a95424ed619690f8e6 100644
--- a/flops-based/gpt2/2/1.pt
+++ b/flops-based/gpt2/2/1.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:00c379fcfd7099cd3eff563dcfc02301dca26064d95900cb2a0245eae68b53e4
-size 242697602
+oid sha256:bd224a80e74b2d920944bcc18a513e6e498fa79bd120e4e3e2de52a19b1d3103
+size 242697738
diff --git a/flops-based/gpt2/2/info.json b/flops-based/gpt2/2/info.json
index 6034600b9912cbad1859998619b92784a9eb0e07..739f7f67279edc0ef82e9bb2faa5aba5e243d175 100644
--- a/flops-based/gpt2/2/info.json
+++ b/flops-based/gpt2/2/info.json
@@ -20,8 +20,8 @@
         778152960
     ],
     "times": [
-        0.012937307357788086,
-        0.00928807258605957
+        0.014228105545043945,
+        0.012090444564819336
     ],
     "params": [
         103174656,
diff --git a/flops-based/gpt2/3/0.pt b/flops-based/gpt2/3/0.pt
index a4e5e74b43454450dc80ac0ba5f6f2dec2d7462d..d764976f6983adc599e410ff395fc0f05f965675 100644
--- a/flops-based/gpt2/3/0.pt
+++ b/flops-based/gpt2/3/0.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dbfa0f2d5212a34df14cfdb6f226ecd9ea9d2af59776d952c1b1381d3488b56b
-size 334130705
+oid sha256:4feaa64327d17762ad0a8fad9f05d4decbbb6a3501fcab5a5f9a3256045336ed
+size 334130865
diff --git a/flops-based/gpt2/3/1.pt b/flops-based/gpt2/3/1.pt
index 43350bc265b1afd9e7981b4d640a35a15ed38389..e4c12f1fbddf4166267eb47978f0cd2eeb6e9df6 100644
--- a/flops-based/gpt2/3/1.pt
+++ b/flops-based/gpt2/3/1.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6e23369b724fb7b01fa2322bd14d148b2b717e8d83ed95293202bb29f2716cfe
-size 176587497
+oid sha256:fbb33507b1ed7d1d96e4d53f24641090041f1ae20f23e5b2764993cc478412f9
+size 176587645
diff --git a/flops-based/gpt2/3/2.pt b/flops-based/gpt2/3/2.pt
index 0795f92e3751f37d9f0f36d58a3612979a3b1dd0..6f794fda6e386178d63a199fed7b214db5ee759b 100644
--- a/flops-based/gpt2/3/2.pt
+++ b/flops-based/gpt2/3/2.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4c276eb6b5314e6479bad5e3796993dda5a53479fc247404ef5f0e6860cca042
-size 154405171
+oid sha256:a7c6b29459ca33d878f5f9bd34eca10b2b4e20b34620fdfb585df030f292691b
+size 154405335
diff --git a/flops-based/gpt2/3/info.json b/flops-based/gpt2/3/info.json
index c8441d032cfe644eda8c9b5d76c7beb872ae7f20..5c2a6b1b7a825c2364c92330b1b73603903ea855 100644
--- a/flops-based/gpt2/3/info.json
+++ b/flops-based/gpt2/3/info.json
@@ -26,9 +26,9 @@
         501815808
     ],
     "times": [
-        0.01198434829711914,
-        0.011710643768310547,
-        0.005055665969848633
+        0.009680032730102539,
+        0.011411428451538086,
+        0.005023956298828125
     ],
     "params": [
         81911040,
diff --git a/flops-based/gpt2/4/0.pt b/flops-based/gpt2/4/0.pt
index af028899bcda5c96d0f8bd0adcd700f9a2873d5d..9fe3be792075aa83f10cd4d6660a251fbc321f9a 100644
--- a/flops-based/gpt2/4/0.pt
+++ b/flops-based/gpt2/4/0.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d8be73ff17e1b185b8c3e24e8c9b3c9f1f7316d06ee0cfa64134c36ff8dc4d14
-size 275269493
+oid sha256:0300334c3bf0b8a87623a0d781f4349b156d9692011b950286b02974238bdd00
+size 275269629
diff --git a/flops-based/gpt2/4/1.pt b/flops-based/gpt2/4/1.pt
index 4da733c16c03c636be35e807d8174d7b6c4c3a86..dbcb55e1f8e5e9fe6d18ed8dbd9c19983ba8c7fa 100644
--- a/flops-based/gpt2/4/1.pt
+++ b/flops-based/gpt2/4/1.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d473eead1092b5db5de79c19a6938029e366ab7eea8feb9a12892eb6d961e3ab
-size 117726285
+oid sha256:10c11a840decb6af4a241acff017cebbad7bf6158651ee78047e43d5bb281c22
+size 117726409
diff --git a/flops-based/gpt2/4/2.pt b/flops-based/gpt2/4/2.pt
index 2f7af329138f2e903e39cf241371a59025b95d6d..a95236eac225db02207ad231452fd48d29f390a9 100644
--- a/flops-based/gpt2/4/2.pt
+++ b/flops-based/gpt2/4/2.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:79b6b0b61fcd1f7b2eac88954b18bb97df45f84fcbdb92d6ec9e6db1a6277f8e
-size 117735729
+oid sha256:010f6ec82e16f75acf74e5ab57fa8a8460df0a5e3979e031227b48b342a19c25
+size 117735929
diff --git a/flops-based/gpt2/4/3.pt b/flops-based/gpt2/4/3.pt
index 239b5d39edf6961d192b2cbae7dc34866b51b067..4d75c9b2402b6f7e2cee92cb3a161e72d3a31932 100644
--- a/flops-based/gpt2/4/3.pt
+++ b/flops-based/gpt2/4/3.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7749c0199d09fc24d8d31d955e36c54817a48af888f699099493b54caff35b28
-size 154395921
+oid sha256:e645795e39feb6af486f184663727246992bf27a189c2ca8ac7ede2d40dd474a
+size 154395945
diff --git a/flops-based/gpt2/4/info.json b/flops-based/gpt2/4/info.json
index 0bdacbd54438e294b20261757a12e6c5e919a53e..c456cc087940b08326f787eb9fb17426d6d5beb9 100644
--- a/flops-based/gpt2/4/info.json
+++ b/flops-based/gpt2/4/info.json
@@ -32,10 +32,10 @@
         501765888
     ],
     "times": [
-        0.00785970687866211,
-        0.008141279220581055,
-        0.007796525955200195,
-        0.004466056823730469
+        0.005810737609863281,
+        0.007517337799072266,
+        0.006907463073730469,
+        0.0046427249908447266
     ],
     "params": [
         67735296,
diff --git a/flops-based/gpt2/5/0.pt b/flops-based/gpt2/5/0.pt
index feaed6a6dfb8991a5305b68f3d55f0f0cdb23078..7a8e89d7ba2e05eaefbecbf8bb603b61614e5fb0 100644
--- a/flops-based/gpt2/5/0.pt
+++ b/flops-based/gpt2/5/0.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bfc90a37b574862d5ceaa690a689885b4f97d4d72f854edfe525f637986896a9
-size 245838951
+oid sha256:6674582c446f71d424b7eed505dbc417967229f13dab026d7153b39acb86e64f
+size 245839011
diff --git a/flops-based/gpt2/5/1.pt b/flops-based/gpt2/5/1.pt
index e62923fd3ea4647d542e3f5d2513a384972182cc..25fb3cbe3157b783b3051692bb16268b51a6981a 100644
--- a/flops-based/gpt2/5/1.pt
+++ b/flops-based/gpt2/5/1.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1a62ec8d8d588fd0e60bdc63e9a015ec845b99f14734c40e80b6630717386230
-size 88295807
+oid sha256:d572548e583454cd22b3024d4c7d546a1c30d6cfa69d22d7eb617d0e8e26a247
+size 88295855
diff --git a/flops-based/gpt2/5/2.pt b/flops-based/gpt2/5/2.pt
index cdb02718af61ee312fb87914e901592c45395e9e..3cbdf2d4d20932bb110ed140880ca47dcd06eb2a 100644
--- a/flops-based/gpt2/5/2.pt
+++ b/flops-based/gpt2/5/2.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d2a8c4ae875931bd29c7ef7bf5a00514b102d852b667651e2369109c815d273b
-size 88295807
+oid sha256:042a199c02a59276e3a99aa1adc79ea62556ab051dc8166a350fc7c896184fab
+size 88295855
diff --git a/flops-based/gpt2/5/3.pt b/flops-based/gpt2/5/3.pt
index 45fc3017905bd3286c89c6ad5e1596d18b0d8bb3..f5d8c2bc46268a52b5f2a61d833ad77a26129c45 100644
--- a/flops-based/gpt2/5/3.pt
+++ b/flops-based/gpt2/5/3.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0857d406db5ff897474ba17761ffa810f51ff39c9424856b4d15783ccb005156
-size 88305187
+oid sha256:739eeb59ee252de5302b387499478a91fca489d19de1695785f943ab08bb2b11
+size 88305311
diff --git a/flops-based/gpt2/5/4.pt b/flops-based/gpt2/5/4.pt
index 6a3272746b28a420f6a3a09b4c60b6d4aba9c15f..9bc1b620967720c0411ec7970af4732aef138f98 100644
--- a/flops-based/gpt2/5/4.pt
+++ b/flops-based/gpt2/5/4.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d684351e0c1d97dffe33868e3ceb9bcdda4bcc34209257982633c34bf2d712b5
-size 154395921
+oid sha256:aefc14f5b41d7c02266f3cb00e51172e374af886bbdbdf6a5324d455e07957e4
+size 154395945
diff --git a/flops-based/gpt2/5/info.json b/flops-based/gpt2/5/info.json
index 70be47a880dedbaf28fa5245380d483a11674d28..202bafde293e4a2fd4e3263e698aa98f95922ddf 100644
--- a/flops-based/gpt2/5/info.json
+++ b/flops-based/gpt2/5/info.json
@@ -38,11 +38,11 @@
         501765888
     ],
     "times": [
-        0.004588127136230469,
-        0.0048944950103759766,
-        0.005957126617431641,
-        0.005729198455810547,
-        0.005186796188354492
+        0.005208730697631836,
+        0.005757808685302734,
+        0.00614166259765625,
+        0.005856513977050781,
+        0.004487514495849609
     ],
     "params": [
         60647424,
diff --git a/flops-based/gpt2/6/0.pt b/flops-based/gpt2/6/0.pt
index 326b9ccca79808ff5f7727f7b758d40a2929e125..ba5a9950ee87a2352980ec34aab34ea6c060e9f0 100644
--- a/flops-based/gpt2/6/0.pt
+++ b/flops-based/gpt2/6/0.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:29c60623014c27b76c1265fc27ed1beb4c0b7f6e72b3b5d3b0b7d7f62fc48b44
-size 216408601
+oid sha256:841ff70216a56af0e875738c5bb27e3f1be5bfaee0fa9427cfec32635bdaddf7
+size 216408649
diff --git a/flops-based/gpt2/6/1.pt b/flops-based/gpt2/6/1.pt
index e99f333c0a559a4ecf0be30d2f0faa13dca88158..59b4a1bec2d0a4a30d0499d4f2e4b9712af76707 100644
--- a/flops-based/gpt2/6/1.pt
+++ b/flops-based/gpt2/6/1.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c8cecf5e6a56c3bb0437e1adcae2aea1f3b5a0702c2595b6b42ce8e0eb5b45ae
-size 58865393
+oid sha256:c17df824a3eb5311581f7e101fcc1739c0e117058e8cf1f0299725979520f2b3
+size 58865493
diff --git a/flops-based/gpt2/6/2.pt b/flops-based/gpt2/6/2.pt
index 997bfe7f5242089eddba403aeed032090d22d406..d20ba9c1aaee40c785d91c65c8ff81ca12c07d53 100644
--- a/flops-based/gpt2/6/2.pt
+++ b/flops-based/gpt2/6/2.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8f9c8545be8edba3c062b91170319370361520c07ede1113656428883949092e
-size 88295807
+oid sha256:75b5c94ced91871013df89b593311ea0a52d513c0b6026414071c269add67cd2
+size 88295855
diff --git a/flops-based/gpt2/6/3.pt b/flops-based/gpt2/6/3.pt
index e1e3e76633def1c548ceb072d9ad0df93b284d34..334e6c98843872de92363bb9cb03193297f2555a 100644
--- a/flops-based/gpt2/6/3.pt
+++ b/flops-based/gpt2/6/3.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0e4b92a58b84cec20ef76a5bf4b3d8c286fb1f6638a59817fb4065269bc86325
-size 88295807
+oid sha256:6fafefa2401c77ac7e6df0da73c8d2122ba0f49eb873d120799ed31310191366
+size 88295855
diff --git a/flops-based/gpt2/6/4.pt b/flops-based/gpt2/6/4.pt
index 49cfc05f63a40d7822399d49f64fb1f95efd1546..c0ec2f09417fa852e0ea6c476c6ed5413813e139 100644
--- a/flops-based/gpt2/6/4.pt
+++ b/flops-based/gpt2/6/4.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b9450367536fb62dc4673a956c7e8c56656aeb50d51606790825a6020150372b
-size 58874837
+oid sha256:e928aa60585e69b12c7b427559fd50b928f4c025a56370750cfaf8261bf1f85a
+size 58874885
diff --git a/flops-based/gpt2/6/5.pt b/flops-based/gpt2/6/5.pt
index 765ee1ceacd0ddff86d821e7a3ff45d790f39881..2d8f388e885217a66f7e47a7a8ebd81e0d8bc45f 100644
--- a/flops-based/gpt2/6/5.pt
+++ b/flops-based/gpt2/6/5.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:072997203eb5e473b58013f724e901ae94a79c128f47ede42153ba4a169f0ff6
-size 154395921
+oid sha256:d805a6074437093714986e66106975eab5d40a7b78c7c06734149d5f59c99e50
+size 154395945
diff --git a/flops-based/gpt2/6/info.json b/flops-based/gpt2/6/info.json
index 198731005b722c880cc2ded7360dd3a747cec80b..b69aa3bad1a3091b4e257e4cfd7eec8de2df64f3 100644
--- a/flops-based/gpt2/6/info.json
+++ b/flops-based/gpt2/6/info.json
@@ -44,12 +44,12 @@
         501765888
     ],
     "times": [
-        0.003908395767211914,
-        0.003949642181396484,
-        0.00627446174621582,
-        0.006484270095825195,
-        0.004202842712402344,
-        0.005176544189453125
+        0.0038237571716308594,
+        0.004636526107788086,
+        0.0045757293701171875,
+        0.004657268524169922,
+        0.003826141357421875,
+        0.004747867584228516
     ],
     "params": [
         53559552,
diff --git a/flops-based/gpt2/7/0.pt b/flops-based/gpt2/7/0.pt
index d19972c179ace5b83ea8d646036348cc56507bf8..bf12cbf9cd80e95ad5b17826df0b3c4dadc52803 100644
--- a/flops-based/gpt2/7/0.pt
+++ b/flops-based/gpt2/7/0.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ea6d27975e6df2025536c847dea59c9c8a6cbb1376416b2988161d51ba3019c4
-size 216408601
+oid sha256:a1248ce13b5d1c9a7d620c13518351364fed1fd49f6fd73b369a5daf8825664f
+size 216408649
diff --git a/flops-based/gpt2/7/1.pt b/flops-based/gpt2/7/1.pt
index 27f446e4c596c1ecbb11fd1fa3a351107560335b..86d9b15ebbaa6765b6db2ff18162550a3f3670f4 100644
--- a/flops-based/gpt2/7/1.pt
+++ b/flops-based/gpt2/7/1.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:60f5a8be0f479af8c81fa57fa1f5440275d2fec6cca02999d8054004dd77d0bf
-size 58865393
+oid sha256:d41465b818b8f34a3dddd62774013ec7c369e3e82c89d3b2e8b95bf8e8746273
+size 58865493
diff --git a/flops-based/gpt2/7/2.pt b/flops-based/gpt2/7/2.pt
index 0d45f1b0294a06dbfba6018c06f61723b075367a..bb34208ab84af8d2e85ded4e7f451bdeaf87b2dc 100644
--- a/flops-based/gpt2/7/2.pt
+++ b/flops-based/gpt2/7/2.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4d2f3b1ccf2717d646eeee28d3bacd9c85910948ffeafba75797e98f75053713
-size 58865393
+oid sha256:2d3106cb68c2610d12c795e334c4a9c1ea0290547a9d0bf64b9cdcc6dc562c43
+size 58865493
diff --git a/flops-based/gpt2/7/3.pt b/flops-based/gpt2/7/3.pt
index 000937fd4d0ce6b0527eadb31cd8f959e7ac5087..9728f14c9d6d44edccd28883dd02a54108dcc491 100644
--- a/flops-based/gpt2/7/3.pt
+++ b/flops-based/gpt2/7/3.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3d7f10da03c445bf4715b8e83b9cfb7833b7d4df11136b8588b1b0f13189553a
-size 58865393
+oid sha256:620cf6d5bbcc468caab3b367f4db9b6045241c3ff37339423deb5f9838f4dfcc
+size 58865493
diff --git a/flops-based/gpt2/7/4.pt b/flops-based/gpt2/7/4.pt
index d2241cdd37cfec0bfe67a7a8da062cd5b32d0881..5921387a4929384d9fcb06f854e4c3332cc43f8c 100644
--- a/flops-based/gpt2/7/4.pt
+++ b/flops-based/gpt2/7/4.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c8d09eb604b84ebd8bf949e35e52e151a89085c736525336568433943c7345e2
-size 58865393
+oid sha256:b21ae7fd166d4b87c6e9ac15a27744aba45e29723359599134192db25f5d321d
+size 58865493
diff --git a/flops-based/gpt2/7/5.pt b/flops-based/gpt2/7/5.pt
index 34296100c36067414469e48cb557de665591fc2a..9c67d9a5b9c755db20d6260a35b96a0d7faacec5 100644
--- a/flops-based/gpt2/7/5.pt
+++ b/flops-based/gpt2/7/5.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:96342b8f98dd2e767685392e0bd1bb0765eb325b48b85a5645e074672a65fb34
-size 58874837
+oid sha256:d95c3e93b5c9818c5b873bec1e3447f79476e2e8685250c3c5e297db4a425449
+size 58874885
diff --git a/flops-based/gpt2/7/6.pt b/flops-based/gpt2/7/6.pt
index 7e3342726095a92690d4cad577d12b729897b817..598efd0ad7d94c363b6787edf531e3e4224f22c2 100644
--- a/flops-based/gpt2/7/6.pt
+++ b/flops-based/gpt2/7/6.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f1b9548ffabff5ff7f3587514f5686e349257f9a6b5c7047033bd7a6169dffe7
-size 154395921
+oid sha256:c2e446a35b708b98575f36884bcbe3884047ba92df63ff361b39e6289b6c4892
+size 154395945
diff --git a/flops-based/gpt2/7/info.json b/flops-based/gpt2/7/info.json
index 04b01084c939d07fd07924caf92d30e728873584..ad395956c59b4f1787ebbfb13a3c928e65c08dea 100644
--- a/flops-based/gpt2/7/info.json
+++ b/flops-based/gpt2/7/info.json
@@ -50,13 +50,13 @@
         501765888
     ],
     "times": [
-        0.003744840621948242,
-        0.0050776004791259766,
-        0.0038187503814697266,
-        0.0077114105224609375,
-        0.005064249038696289,
-        0.0036427974700927734,
-        0.004510641098022461
+        0.0034155845642089844,
+        0.0046579837799072266,
+        0.003651142120361328,
+        0.003980159759521484,
+        0.004025697708129883,
+        0.005561351776123047,
+        0.0044744014739990234
     ],
     "params": [
         53559552,
diff --git a/flops-based/gpt2/8/0.pt b/flops-based/gpt2/8/0.pt
index 04709f277159ebef5b04c787de30f488f924bb9d..7b93200556da84365eea9924d64bf4ed93d27b5d 100644
--- a/flops-based/gpt2/8/0.pt
+++ b/flops-based/gpt2/8/0.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0eaed34b4a828997ce189b798138ecfa0deab8a103e35b3fbfa303d2a9bb63dd
-size 186978251
+oid sha256:b25cbd1372d6775a4d2015bdc457f74a79a74c43027d37164004898d5821dbfb
+size 186978287
diff --git a/flops-based/gpt2/8/1.pt b/flops-based/gpt2/8/1.pt
index e8cced6e790a77fe7ed55c4727eb0bb3a2549263..d04df8f2f605fd26998b032bc8a242fb8b71da01 100644
--- a/flops-based/gpt2/8/1.pt
+++ b/flops-based/gpt2/8/1.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f92d6ea077941d8fa70b76bc15b6a2ac7f70c4e66f52fe825e1929d7a3925063
-size 58865393
+oid sha256:102d2f0d7faed9e422bc1e3d931d44fdf57f29e4f139f117effcabf00b7dfca3
+size 58865493
diff --git a/flops-based/gpt2/8/2.pt b/flops-based/gpt2/8/2.pt
index 4f7a8f780b0bee56fc910b8db8ea9baa2d0e5ac5..204599291a1e43a1ffec3ea9ca853ffd955dfd9f 100644
--- a/flops-based/gpt2/8/2.pt
+++ b/flops-based/gpt2/8/2.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4625d398a4cb8fe149d902344cb054de12d49afa36ae6fc46bad4884c702521c
-size 58865393
+oid sha256:3e024cfb3fafa62bb728b1842319d3954ef6d4001c3d488d9ad0f57e77eabc27
+size 58865493
diff --git a/flops-based/gpt2/8/3.pt b/flops-based/gpt2/8/3.pt
index 9e7504bf3a6adb30c32baf525566cca36da984b5..9af5055f90f62f909e074f336f8ebf72871cd86f 100644
--- a/flops-based/gpt2/8/3.pt
+++ b/flops-based/gpt2/8/3.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:925820fe58eca35ed2d2053da5a5da8b11fc5f5c3effa096cca129211969ccac
-size 58865393
+oid sha256:45f0ed0cb0a262b0a68b76711d78cd5c43676f95ca041cbed8775ea3e47bebe3
+size 58865493
diff --git a/flops-based/gpt2/8/4.pt b/flops-based/gpt2/8/4.pt
index 145cd85f771a42a8e33737ebf737c1c3c032e70f..12c929d3c8b3dffd5107d9b79ac80b73e79a6b0b 100644
--- a/flops-based/gpt2/8/4.pt
+++ b/flops-based/gpt2/8/4.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7038c71dfd7c436d0e38aff436609d859b2107afa35bf9a15b2dadec9a07d61d
-size 58865393
+oid sha256:5fc20eb876af3aba16897aca4f1960747187ecc37ff67c986c406f9497ecee29
+size 58865493
diff --git a/flops-based/gpt2/8/5.pt b/flops-based/gpt2/8/5.pt
index 0f93481dd5ad55154d1950d7020b8bd9b108ad42..c973018c73673ee86de412cc632a65aae7714bd8 100644
--- a/flops-based/gpt2/8/5.pt
+++ b/flops-based/gpt2/8/5.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9fd584f66338d7803a609c45e454474276e1f57d33fc09121406937fc03c23bc
-size 58865393
+oid sha256:37011e66feb3ef65acbe6228b95f61e0508022dc06b7db1608c69ce220489bd5
+size 58865493
diff --git a/flops-based/gpt2/8/6.pt b/flops-based/gpt2/8/6.pt
index a4ae25b3801e4ec82ed4a126f79ae4400d4f99f0..73f30873ecebde6b4efebef5e4f59526ad37170c 100644
--- a/flops-based/gpt2/8/6.pt
+++ b/flops-based/gpt2/8/6.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8d156416d398431592aea36cbc044e8c036aa9e74f8eb9feeda354178dd4b07a
-size 29444359
+oid sha256:dddff8a48436789d1ce7ef922c3b0fb9b3126ffd692c198038f39420d732baab
+size 29444523
diff --git a/flops-based/gpt2/8/7.pt b/flops-based/gpt2/8/7.pt
index 4851d8aa40df4af8c94b6e4cc48b025890d44273..525b27506aadc90d578377986f5499b9b2504fb4 100644
--- a/flops-based/gpt2/8/7.pt
+++ b/flops-based/gpt2/8/7.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:758e9878fea181e1632569ddb7862943bd8c53b0dc195b765ae6a0100eae1190
-size 154395921
+oid sha256:5dc01f4dd309849cb2a7fbf367b1db74b02418adc0f52f1cd877b04f4336bb40
+size 154395945
diff --git a/flops-based/gpt2/8/info.json b/flops-based/gpt2/8/info.json
index ad2c409d01169a176383a33843b672189da89d86..5d34c1bdeb03d5c347d5d85e5d7a48601b409c30 100644
--- a/flops-based/gpt2/8/info.json
+++ b/flops-based/gpt2/8/info.json
@@ -56,14 +56,14 @@
         501765888
     ],
     "times": [
-        0.009957075119018555,
-        0.004663944244384766,
-        0.004091978073120117,
-        0.003492593765258789,
-        0.0035479068756103516,
-        0.0033326148986816406,
-        0.004043102264404297,
-        0.0074024200439453125
+        0.0025146007537841797,
+        0.004233360290527344,
+        0.0049517154693603516,
+        0.0041806697845458984,
+        0.003915309906005859,
+        0.0038356781005859375,
+        0.0019066333770751953,
+        0.004812002182006836
     ],
     "params": [
         46471680,
diff --git a/flops-based/gpt2/9/0.pt b/flops-based/gpt2/9/0.pt
index f414bb46d01994abb56b1b01fc043724c29f1375..8fe8fb1b6ef4a11047e9a995921d9955bf54fc51 100644
--- a/flops-based/gpt2/9/0.pt
+++ b/flops-based/gpt2/9/0.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:608ec30ace7006232d07dc828136df55731ce71e7ed5cc6fd66fc5e900d6d53e
-size 186978251
+oid sha256:3e6ab798cdf89bd919b41534fc30a8db4cf57d60a3b9c405ff13d26c2f29af19
+size 186978287
diff --git a/flops-based/gpt2/9/1.pt b/flops-based/gpt2/9/1.pt
index f2fe082e86e28e65005cebf7adfa8da69a02ed35..a6f0590def02569a9f4450645c501b8a2af74fde 100644
--- a/flops-based/gpt2/9/1.pt
+++ b/flops-based/gpt2/9/1.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e8851b95bfb41136f4cf23afdb8f940a12f7f99e62d4b96ac6e873abd2975da6
-size 29434979
+oid sha256:04e26de9ea7dcb25f7b95b64715a9e0a7778bc3da8ff5725caa8e9ad431ddc49
+size 29435067
diff --git a/flops-based/gpt2/9/2.pt b/flops-based/gpt2/9/2.pt
index 2830dbbdbd92cf5f51a8bfb6d8e6f6aadbb73ece..6aa7d01689e13cecad35e6f1dc60700cd7952c80 100644
--- a/flops-based/gpt2/9/2.pt
+++ b/flops-based/gpt2/9/2.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9c9693bd20f2725dfa71260538e1c0989041cf6b258117c1f767f423b4b5be8d
-size 29434979
+oid sha256:b72bdffa2b72c324bbb6c3f290e776f072ba3f5c79bc21b269c56fa36c7e76c3
+size 29435067
diff --git a/flops-based/gpt2/9/3.pt b/flops-based/gpt2/9/3.pt
index 079e502b9f0bc0b740d0abf8cf96a416a3f310a1..c5c0a8740b8d28cbcc6b7b25145fe744f5460195 100644
--- a/flops-based/gpt2/9/3.pt
+++ b/flops-based/gpt2/9/3.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c149e6b2ca7d2cb172c304b577b76c354aef6712732c2d54b119f3e7fb6dd394
-size 58865393
+oid sha256:c6fbdc21d0b494a55545a3510d77c1984a3c21ded7644520ba146d82f2ff6852
+size 58865493
diff --git a/flops-based/gpt2/9/4.pt b/flops-based/gpt2/9/4.pt
index 2c002e84627ce0bdf4bb619c1a666303ea27d3b9..65892717f6ec939747ba61fa8bd1f8a129a78688 100644
--- a/flops-based/gpt2/9/4.pt
+++ b/flops-based/gpt2/9/4.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d47625b0ff1f02d9260aa60301ccd59a113af3f10ca7ae0c1d88d7d26ae1dbfb
-size 58865393
+oid sha256:3d10682b3bcd75c2e86f971b0315c6aef8e10c02103c2c2309dcb29acfca5bd6
+size 58865493
diff --git a/flops-based/gpt2/9/5.pt b/flops-based/gpt2/9/5.pt
index 8a5a802e5eaa10ac2d2bdf191a6986de167d0f35..a9530bd8d75045aee83e558c151fc74da95098a1 100644
--- a/flops-based/gpt2/9/5.pt
+++ b/flops-based/gpt2/9/5.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:619d554fe02ae767d5dead1f13506bf1abf3e37ecb7d8f3860489db290c4e20a
-size 58865393
+oid sha256:35e208d811b3d0178137fc692f975b9cd199fc4ac8002579d4c21534fbc34ef1
+size 58865493
diff --git a/flops-based/gpt2/9/6.pt b/flops-based/gpt2/9/6.pt
index e0f27037eabc099e6661b8d1424512631af334f5..22e59f1ec675f47757162aa84779b55de672283c 100644
--- a/flops-based/gpt2/9/6.pt
+++ b/flops-based/gpt2/9/6.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:750858eef15af46e2a71348c4a9618b39de73246b1072f47266b6df6169bc0e2
-size 58865393
+oid sha256:0ae8c0de8a4076b5621e1df894237896b1635fe73295edc493e9d62a2c60316c
+size 58865493
diff --git a/flops-based/gpt2/9/7.pt b/flops-based/gpt2/9/7.pt
index 8e14a82a5484569bcb231f83ed8fbf9bf511c236..e2a2fc4e8f009cfc0f14852d61ee8444729166d9 100644
--- a/flops-based/gpt2/9/7.pt
+++ b/flops-based/gpt2/9/7.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d0a7c1c435f5968fb8ac89921f2667abc5919beb23e2751bdf23d7750d9f030b
-size 29444359
+oid sha256:b300388569a299fa3e402a9fd6d8fb3359ca8af02fffd256756ac336a69b6827
+size 29444523
diff --git a/flops-based/gpt2/9/8.pt b/flops-based/gpt2/9/8.pt
index e3ccf71250223f04477709e0a45f862f63a208cd..a28f59e099c7c66e48a045e00f87e863254fa2ca 100644
--- a/flops-based/gpt2/9/8.pt
+++ b/flops-based/gpt2/9/8.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2c561b699dda33441fe3a7f2670914f19189f0c20735ccfa84a2692335f5c681
-size 154395921
+oid sha256:5285f113a6dbdcf7a84095e9424e74eb5f3919879aeb2dad7bf009c22f70c3be
+size 154395945
diff --git a/flops-based/gpt2/9/info.json b/flops-based/gpt2/9/info.json
index 44505b8b80ef730633c894c7c8dd73fbed9de05b..474300bb5f5768b73235895b3bd4bba57b057259 100644
--- a/flops-based/gpt2/9/info.json
+++ b/flops-based/gpt2/9/info.json
@@ -62,15 +62,15 @@
         501765888
     ],
     "times": [
-        0.0026247501373291016,
-        0.002599477767944336,
-        0.0024156570434570312,
-        0.0035800933837890625,
-        0.004146575927734375,
-        0.0038535594940185547,
-        0.003537416458129883,
-        0.0019214153289794922,
-        0.004603385925292969
+        0.004093170166015625,
+        0.0020084381103515625,
+        0.001920461654663086,
+        0.005430698394775391,
+        0.0036170482635498047,
+        0.004530429840087891,
+        0.00299835205078125,
+        0.0017931461334228516,
+        0.004969120025634766
     ],
     "params": [
         46471680,