diff --git a/flops-based/gpt2/1/0.pt b/flops-based/gpt2/1/0.pt
index 010e173fdfdbbe14c52218695499ef89e4bdfb41..6af93d370dba784ad437fb369d005d3094c8341f 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 sha256:8c304fd5feac770deddbdd9993b274cf21a5c9bc47a9fdea70df752a9d5124e3
-size 510726905
+oid sha256:7bd55abc33779ea131ba879acd582f613e21f37412d230a33412c829fa615ed8
+size 510726333
diff --git a/flops-based/gpt2/1/info.json b/flops-based/gpt2/1/info.json
index b6cdc240d6151147e195fdb3d6a3c39f7f083112..3c8ec8137f54f657053d8da527e3ad5ceee23bd6 100644
--- a/flops-based/gpt2/1/info.json
+++ b/flops-based/gpt2/1/info.json
@@ -2,11 +2,11 @@
     "shapes": [
         [
             1,
-            13
+            0
         ],
         [
             1,
-            13,
+            0,
             50257
         ]
     ],
@@ -14,7 +14,7 @@
         1607164416
     ],
     "times": [
-        0.022766590118408203
+        0.027774572372436523
     ],
     "params": [
         124439808
diff --git a/flops-based/gpt2/10/0.pt b/flops-based/gpt2/10/0.pt
index 70176fd3e1f2bdf1c750950d6dc9c51961c3f770..7eae78d465dc28069987074f16d11b258b54d639 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:ec0fd6b746c9ff21a8762f5e0530b96623e2d91deb24e53f42ac5b4f3bd2212c
-size 186978287
+oid sha256:d368b045790587c3e427160aae6ba8af8d73b6cbb8eb92c916966eb24fdcf59b
+size 186978245
diff --git a/flops-based/gpt2/10/1.pt b/flops-based/gpt2/10/1.pt
index 2b896205c0e4d40601c94f62cd8af6b9602795fe..27e4d7e19d515c740464b86022a1058b27900664 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 sha256:78bf9c89d480ec74c3034d974d46409e642a893eb35855ba4f448a78fb275221
-size 29435067
+oid sha256:dc15a2896f446395bcb8975379b1a5327bd55f35024d53b0effd0626b8b8bcd0
+size 29434969
diff --git a/flops-based/gpt2/10/2.pt b/flops-based/gpt2/10/2.pt
index 872a3ca9077bf5703ea446eb8818cf11673d3d2b..b67ea9841e67d7c63fb767be34fd89c3a8b3c712 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 sha256:592bedc00952f4a94a456bc5c9ab320130af298fd9aa043d396d418d7a51b972
-size 29435067
+oid sha256:ea89dab6064497455bc05c33980af2da4ade0c6ac2b2150bde3f49dbc88eacf2
+size 29434969
diff --git a/flops-based/gpt2/10/3.pt b/flops-based/gpt2/10/3.pt
index 03599f77a46518a8e2176124f3c7903f94efbab9..897dc97f4739c4145dab305699a67518b274fae7 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:dab95015637342d81c946b92dcf69b76d0b1a80241bce016a22d10e9c19627c2
-size 29435131
+oid sha256:b7db30d33696b623a6699591c5bdfb6c77dd441531bef24ab834726ead6b8f2a
+size 29434969
diff --git a/flops-based/gpt2/10/4.pt b/flops-based/gpt2/10/4.pt
index daa20f822d1d183365620f7db0ad740fd3daa483..9d98f0cd6c022f6cc0ff198e072bd2f04d4bc52a 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 sha256:5b5b138d60e0dcf1e47c0a0dc30464ca983b19e4298f1a7390f735914d54800d
-size 29435067
+oid sha256:e793788dc48defa2e75bd6cffcf0e611494fea77a606b126b8dd6447346a151e
+size 29434969
diff --git a/flops-based/gpt2/10/5.pt b/flops-based/gpt2/10/5.pt
index b699b55844e685732c2aa7746c3b2aba1e76f6d4..c72614baa76074e4babd07444632266f444de7f6 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:dadcae09fb03caad991674b65d0c815b73a76a8f454ffb815ae90800ade9ae5c
-size 58865557
+oid sha256:d9f6725007b1a7ab2a0167f0705b30006c9a2a8db4e183542f7b93669f4114a3
+size 58865365
diff --git a/flops-based/gpt2/10/6.pt b/flops-based/gpt2/10/6.pt
index c449d87b9a61ce12a2e08002bc6bfa1fd4eaddf4..cc614c55f8a85315263869a4cb630b0527fb6b28 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 sha256:29ee4bd1b483941a829fb1434f7ed747358bf46cf2a330a92eddc0d6497fdb03
-size 58865493
+oid sha256:0803d4c925897e54839eed646606955d8b3882dcd738f28709790cfef56d51f0
+size 58865365
diff --git a/flops-based/gpt2/10/7.pt b/flops-based/gpt2/10/7.pt
index 67f162ee4f04f9b42eb717f90e1c7503dd046212..e0cf196cac2077e610e4f76e248305a81d9a1768 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:e194ec94e2046730f6b65b6ce707fe474d3d721ae87b8881401d1ac604a7371f
-size 58865493
+oid sha256:474ea8b66f05808d90b4fd2e0d193bf2b65b93643175b19f022a6fb3fa4efa80
+size 58865365
diff --git a/flops-based/gpt2/10/8.pt b/flops-based/gpt2/10/8.pt
index 02d64b823b792d900d49755666424fcc0183b4a5..de4097a34cb07be3986f2c09ba5e50ca3089665d 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:f80c3143fa7e74a0702d806c470df228290793671342ea4d67cbed3ffe50bda5
-size 29444523
+oid sha256:45b705800b95a10b07db0d701d34ed83d67c39f9068b8796810cf6a4c3d27b9c
+size 29444485
diff --git a/flops-based/gpt2/10/9.pt b/flops-based/gpt2/10/9.pt
index bf8988145f5e5c7e1d36df747b0628aeb3cdb62e..f7bd7546e1cc3d8c46bb334ebefde92439952d90 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:d3df76c2b6db56fbca7ffe8a60992cb791c22b203eaa28ca507c2aaf253bd772
-size 154395945
+oid sha256:db2cdf6f33e33634253f016b64d65b1faa8767d468c14ea2d53e926822c1603e
+size 154395937
diff --git a/flops-based/gpt2/10/info.json b/flops-based/gpt2/10/info.json
index 4cab6f975b0a9bc75197c4822de7e84c400736f6..f87b64aea937d3fc36dcfdd61efa6657ff0c070f 100644
--- a/flops-based/gpt2/10/info.json
+++ b/flops-based/gpt2/10/info.json
@@ -2,56 +2,56 @@
     "shapes": [
         [
             1,
-            13
+            0
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             50257
         ]
     ],
@@ -68,16 +68,16 @@
         501765888
     ],
     "times": [
-        0.0023927688598632812,
-        0.0019953250885009766,
-        0.0022687911987304688,
-        0.00250244140625,
-        0.002256155014038086,
-        0.0032486915588378906,
-        0.004351377487182617,
-        0.003931522369384766,
-        0.002094745635986328,
-        0.005228519439697266
+        0.0036592483520507812,
+        0.002232074737548828,
+        0.0021805763244628906,
+        0.002511262893676758,
+        0.003760099411010742,
+        0.0043675899505615234,
+        0.0040361881256103516,
+        0.0051190853118896484,
+        0.001964569091796875,
+        0.0050182342529296875
     ],
     "params": [
         46471680,
diff --git a/flops-based/gpt2/11/0.pt b/flops-based/gpt2/11/0.pt
index 8a5ad00b9929b9278c620cbbf697d3dfe04faadc..72e1997b8d0696d8745af6e0f3c1209978ad2b29 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 sha256:87fd1f2326ea0aff3d4fd5b5585c9e0ffbe10ad9d77a6ba43bdfc9767c091d0f
-size 186978287
+oid sha256:b043be964b72c061bf91295e372f247ad453dc6d52be7c44f5a2de14c94d38bf
+size 186978245
diff --git a/flops-based/gpt2/11/1.pt b/flops-based/gpt2/11/1.pt
index d1239dac5bb4b0aa48c6e72c860ef09d36fd5cf2..e32922cdd1a1766166ef544020f1c3e1cde1a3f3 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 sha256:5d56fc7b5a26673ac514ab72e8712692847f314b9cb0b7e30ad0c1c9f8e4668d
-size 29435067
+oid sha256:c713b504b6e4f4954b44fb84c47c818f2e6494b087944b7e1ae9eaef7cb1b004
+size 29434969
diff --git a/flops-based/gpt2/11/10.pt b/flops-based/gpt2/11/10.pt
index 17df5393e6c6a7538e5d7004c11347aca330ad87..197ef2902309533eee8b57fbbd563707b182c128 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:e478302f030cef5255cfcdcec188ff148167b8caebec7cf80dc0d2a951322d6a
-size 154395961
+oid sha256:c7315b045857fb87e9ed80d3ad833b0b0265d45758dc048c3e1aaeb23917d7d3
+size 154395953
diff --git a/flops-based/gpt2/11/2.pt b/flops-based/gpt2/11/2.pt
index bb14e9aeda1e04ed6c501664f516e11dca42ce0b..e57860ad117244bb6f5fbddc8715846b24dad611 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 sha256:87345183eae9665ee48f676c2c06c8f0502ab5cdea65aea167d0f586c3f134ea
-size 29435067
+oid sha256:ebb944113353363a3020ec30e27203541cfa49550c008b371119e2ab58f067a8
+size 29434969
diff --git a/flops-based/gpt2/11/3.pt b/flops-based/gpt2/11/3.pt
index b4bde2324262af45620da515eb9d1c81e5d05f1a..09355b1615534bb1ebca734e02fa2a327e8dd901 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 sha256:8febf8c44176801783d647af69b877a36a4a8511999d06f93d2a790b6c48d2e4
-size 29435067
+oid sha256:91ca9d04e6010430dbba3d6fb430ccd5d842d88e8b4d7e9a5acc19899dd60d88
+size 29434969
diff --git a/flops-based/gpt2/11/4.pt b/flops-based/gpt2/11/4.pt
index df308bcbe744e6f44d270d7d08ea3e02768a4cf0..0032d137a3e7f8d97f60f34f23c1ec9c68bd003d 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:a47a6aca3c0feab5e3b2306d0c00e31405e94d0e7fafd6fd2c9aef430d5e0fd5
-size 29435067
+oid sha256:1d01202308b141d04462e7424fc988fd8efeec238258f97f8b6dc7468c19b24b
+size 29434969
diff --git a/flops-based/gpt2/11/5.pt b/flops-based/gpt2/11/5.pt
index 1cb48c630b22c174f76841488662028ea047e786..fb6c077666b7ad66792d1377e719ffbc3b8d2b9a 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 sha256:6b827b9d4e1f1da385c1a89fb9d4d0abb09a1814f40f22ab64addaef838bb04b
-size 29435067
+oid sha256:753315abf8920a46439e1476c4ca01e9dcb4270428f272f201d61e4111ed984b
+size 29434969
diff --git a/flops-based/gpt2/11/6.pt b/flops-based/gpt2/11/6.pt
index 8c413aea347e8827e826e0e27cb6c15d81b4b63b..5477fd0b703e450e9f46bbc522deeaa4acd48161 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:c7aa3f815f9f6d1a20a76248a113dcbddb439a128c7dc6210e12d25897616039
-size 29435067
+oid sha256:72ceb332ff3b2d5358169ebe0a876f785b3340a71a488cd95ae6f93fc46dc69e
+size 29434969
diff --git a/flops-based/gpt2/11/7.pt b/flops-based/gpt2/11/7.pt
index c01390baf717dac88bcdf7c58ff3e940dff686bf..c7cf8b95e9912e89cf2d958176edb59b397f8a32 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 sha256:679ae435b52b258232864f15b558c66210dd968fd017097164a76822af30543e
-size 58865493
+oid sha256:08a9048511323c4f4043f205950fc00dc23f78ddf5c74a2a82352c105212eb6f
+size 58865365
diff --git a/flops-based/gpt2/11/8.pt b/flops-based/gpt2/11/8.pt
index 2eae727e5e461bcc12eebebb93493005d7eb62d9..7626d6834969a5f58589b1b006b3a0fd1c6aa898 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:cbead182734f5020d831f633df884aeea695e60c38f8a0782ef3feacd21f52ed
-size 58865493
+oid sha256:fe8fd0647b2386c4bf2e02688bbbcce78d54e62701673d9cb52730c0e0412807
+size 58865365
diff --git a/flops-based/gpt2/11/9.pt b/flops-based/gpt2/11/9.pt
index f108b0af950018bc8079b67061ef50373acb1cb2..7bf36ab549ad70dc66f92cab290138f575813979 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 sha256:0e0ba1fab6a2edf8625f1538452693a359112dfb3bc00059392792011bf5d99b
-size 29444523
+oid sha256:84492c0b38e380cb05e54f57e93f1cc80e42dae9cbc98ec018fa2498ace57147
+size 29444485
diff --git a/flops-based/gpt2/11/info.json b/flops-based/gpt2/11/info.json
index 27675c8684bf0f0c4158d725e816cbc413320df1..3199377c013a9fd9b92fe31284d2cb95fd8b70be 100644
--- a/flops-based/gpt2/11/info.json
+++ b/flops-based/gpt2/11/info.json
@@ -2,61 +2,61 @@
     "shapes": [
         [
             1,
-            13
+            0
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             50257
         ]
     ],
@@ -74,17 +74,17 @@
         501765888
     ],
     "times": [
-        0.0024166107177734375,
-        0.003831624984741211,
-        0.003988742828369141,
-        0.00365447998046875,
-        0.0035195350646972656,
-        0.002308368682861328,
-        0.003462553024291992,
-        0.0036993026733398438,
-        0.0059032440185546875,
-        0.0023107528686523438,
-        0.004665374755859375
+        0.0033578872680664062,
+        0.002077341079711914,
+        0.00408482551574707,
+        0.003938913345336914,
+        0.0026912689208984375,
+        0.002824068069458008,
+        0.006006717681884766,
+        0.007441997528076172,
+        0.0038411617279052734,
+        0.0032753944396972656,
+        0.00491642951965332
     ],
     "params": [
         46471680,
diff --git a/flops-based/gpt2/12/0.pt b/flops-based/gpt2/12/0.pt
index 5293ad83bd03e8c4c490e2e7055d918915ef6ce1..d7a257142d2950df300db0c2f6adad9982fd2d05 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 sha256:9f6bb3fc75e04e125f3fa119614ad1f53bdd62073a69386b3c86ef602de722fb
-size 186978287
+oid sha256:a78aa71cdc3c84fb8ba3b8152db407688b6df1a6174ccb421e71a4e5bd956f6e
+size 186978261
diff --git a/flops-based/gpt2/12/1.pt b/flops-based/gpt2/12/1.pt
index 46f9a85f4c6fdd2e4aebc3c433ed553e998d1711..206890ac3abe85329289930eaf124de45ae4b368 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 sha256:86da82964f62e1ad26d021286450756788e05f09ed58fbadca1b46a3dffc8ce7
+oid sha256:e5013a496b3a1055f4b2d6b497fe801bc31df6714aba74885e9d205039066828
 size 29435067
diff --git a/flops-based/gpt2/12/10.pt b/flops-based/gpt2/12/10.pt
index 5987071c843370d8b0825058ec0537dbdb2dbce9..b4e066ccdc6c5cdcf223ef47a35b3caf236de7e1 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:c42b911603bb3f21a8d949e48e1b4f9e718f8730b34448cec9d903cf0593b9b6
+oid sha256:0a85e7d6981e42ce38d834bf9a1ca7d1e29dc442ce94d3b83338f099c9a4d369
 size 29444716
diff --git a/flops-based/gpt2/12/11.pt b/flops-based/gpt2/12/11.pt
index a2333058be2a1aa6f7403cf76c857eae1c1867c4..1fbfb7f270afc8a3c9d7975fc0f855b432928a61 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 sha256:4a090e75c31929ab85bbdc9ffdc2e8f5d397266fe596659b46d972c5274a743c
+oid sha256:3d4d797efa85e4b99da341bd71111887942c64293ff62abb9472699ed8858dad
 size 154395961
diff --git a/flops-based/gpt2/12/2.pt b/flops-based/gpt2/12/2.pt
index 62ce58c01e14124dbe737103ae9307a41b2dc03b..9214f6e22f72e569cbe6d7f236f0e7064edba5b0 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 sha256:0d49d76a0264bf7dadd8abb025e48087ad60e8ed513adca56eac457aa0ceb74e
+oid sha256:6d40149c913ca6ce8c11b8f8cc7e48b8dfcd9f6d0745fd3fc2d4f03c9781d900
 size 29435067
diff --git a/flops-based/gpt2/12/3.pt b/flops-based/gpt2/12/3.pt
index 1ec6ec30e93696a14781c8b3cd3262ce71d92321..af2a59da6b4f41eded78ded613a4cc4919f86bb7 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 sha256:90b2a73767b9ebcd8718e57cafb4c1c09287d3b0e122b6870b11fd0240985d63
+oid sha256:72806297b39fdd3516b4e19204f74a60a84442a03876e9aa6df93da1d418f829
 size 29435067
diff --git a/flops-based/gpt2/12/4.pt b/flops-based/gpt2/12/4.pt
index 41223e6551704e1865a4c4d4c16b5220f9c84098..943e6b8204c3c1647a79a11b1b932dce23e5d0ab 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 sha256:85697de35ba8e100e8712fb851e24fa3ad772e6510dac009360bc27d747d6012
+oid sha256:da06b26a75b5d13e7af95bfbdbd84a5321f32f226b9bc5f212a5ec54e71f53cf
 size 29435067
diff --git a/flops-based/gpt2/12/5.pt b/flops-based/gpt2/12/5.pt
index fc137ca44856044a904cf5f89a0ba96c06defcc4..1e62efa0913d248d7de22643debe13cc95e4d17d 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 sha256:559948526b479f93575d248b991724c4c8249e7cb88ddec9aa20aff7626fecd0
+oid sha256:8796c75f4f7cdffb0b4fd26e468e41d5d1edda8ebf85c24ab22967f6414a6cc9
 size 29435067
diff --git a/flops-based/gpt2/12/6.pt b/flops-based/gpt2/12/6.pt
index 51b1a63b8882996f708e486ea3722f481c663682..a4c0df782bb59678303cfb344a03267f4cddc49d 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 sha256:36fec45fd089b3660ed0a308f021429ed4be83889af31c93fd78c0375e457211
+oid sha256:0340daf5a6afa3504150140ec922862a9c594e0bf7553318b87849bd10b8e37e
 size 29435067
diff --git a/flops-based/gpt2/12/7.pt b/flops-based/gpt2/12/7.pt
index 1bcb6edda0e119dcc406eaf942c5fd579514c0f6..10d113f2c98d33bcefff1e50d2dee817e90215a3 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 sha256:1576718ccf4be4aa4cfa275ca28bfc84d4c2bcc417aa58b59550bdaf146e5dfd
+oid sha256:3bb7490ead02d92e9691f98c741c4e97002a48bbb2ccce5606bc9cd9fa17d969
 size 29435067
diff --git a/flops-based/gpt2/12/8.pt b/flops-based/gpt2/12/8.pt
index 09f1e38289c4236f962ee49c23216b5cde86fc11..19c577880ab3ef158a565fca0f404a2809d39615 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:d09e4b1f452d3716f2517915d299787e8f89e75bbff93d7e10ffd1b24fffaad7
+oid sha256:29d73502dc4166ef86bc729c7edb9398f709d8883b563c3cf121f53a5e8d4c94
 size 29435067
diff --git a/flops-based/gpt2/12/9.pt b/flops-based/gpt2/12/9.pt
index 342676a082b3ea2c0fea7a1c1a824a8e3efd7f1a..ccf17d62a75315ae65fc349269c2e56d596b1584 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 sha256:9bfc4fe5421abc9a29b8cde16e356547adf1f6bb5f0141333c8302549a230f6e
+oid sha256:23c2bf9b157c50293726289732b8471cd5d5a10a1790e728bb2a30c052898534
 size 58865493
diff --git a/flops-based/gpt2/12/info.json b/flops-based/gpt2/12/info.json
index 51e0735bc50bc98510d8f5f9ec4f80cbdc3384c8..a64e4daa002c770b715e0f171e281c7bc88d72d7 100644
--- a/flops-based/gpt2/12/info.json
+++ b/flops-based/gpt2/12/info.json
@@ -2,66 +2,66 @@
     "shapes": [
         [
             1,
-            13
+            0
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             50257
         ]
     ],
@@ -80,18 +80,18 @@
         501765888
     ],
     "times": [
-        0.002263307571411133,
-        0.008869171142578125,
-        0.002043485641479492,
-        0.0037772655487060547,
-        0.0022172927856445312,
-        0.004255533218383789,
-        0.002051830291748047,
-        0.0020220279693603516,
-        0.0025339126586914062,
-        0.0047757625579833984,
-        0.002246379852294922,
-        0.004633665084838867
+        0.0030508041381835938,
+        0.002787351608276367,
+        0.0022568702697753906,
+        0.002423524856567383,
+        0.007197856903076172,
+        0.004573345184326172,
+        0.0027239322662353516,
+        0.0025191307067871094,
+        0.005039691925048828,
+        0.004331827163696289,
+        0.0021123886108398438,
+        0.00480198860168457
     ],
     "params": [
         46471680,
diff --git a/flops-based/gpt2/13/0.pt b/flops-based/gpt2/13/0.pt
index 37665e361729ca31a5c3a0944c228cfa85a0e27d..82fbb7633306a7cfc5efcfced36b4a048b25a307 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 sha256:224d313091361a98b92b181b91fcda9dd12dc35407d4cefaeda7d9a0d4ec3b30
+oid sha256:14c8c7f17a963344db0edf8d2bc7f8d61a7bf4af981f0f527cdad100bd03af71
 size 186978287
diff --git a/flops-based/gpt2/13/1.pt b/flops-based/gpt2/13/1.pt
index 3aab150cf04826478608c36e7cdc87e39b1fbdcd..8852339b9e15e74f958866da557f25fedfdff5a0 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:b81e83da135dc8b0becfc45761fa7ebb76aea11cdb8030132b35e99fdf93067b
+oid sha256:374a4e6fb3d3ae3f8a5919201524d2058d105d7371a1d850156db22d93fe542e
 size 29435067
diff --git a/flops-based/gpt2/13/10.pt b/flops-based/gpt2/13/10.pt
index 593fde3660d1f06facacb1afb70581f65100916e..5755e21d85c1bbb55e22963763cebcf1128dfb44 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 sha256:4ece51cef72c34ef85dd2ec0877baac0f353927d60ba6bbe31653cb69ca39d19
-size 29435254
+oid sha256:fe3b8db1e5baba041156c40294c3c5d7257a23a23d3ceba19f6526669c6ece79
+size 29435190
diff --git a/flops-based/gpt2/13/11.pt b/flops-based/gpt2/13/11.pt
index 8a134825d6f3981714a7f350697094e4a7e6c43a..9a2e2c5a5a469831b322b6e57d0ad5e4f4a51c98 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:b9674a2601e65c3b0715329245c710131951741129cd1e33b250b0966e38fe66
-size 29444652
+oid sha256:99aa6aa5c308aaf2151fe402da44d376c5ee363626f984e93ac831afdcd3e145
+size 29444588
diff --git a/flops-based/gpt2/13/12.pt b/flops-based/gpt2/13/12.pt
index 9dc25c948a04b50df8af72636a1755672651bec4..54a8bb5c8ceff10373e195282d3e98a1bbcf9755 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:a6b0c5c1ac94c0845c458b8be9273534531be7274eb9a0d3c89a6c3cd2104584
+oid sha256:42cccc93202cfe83db3da1231b563ce884f703ccbb51481ae572e0b2dcf87a19
 size 154395961
diff --git a/flops-based/gpt2/13/2.pt b/flops-based/gpt2/13/2.pt
index 96a9aae1aadc033a192c84e5438a816268c7d7b6..885a0368714bbdb5029cfec16e26f75b66e7373a 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 sha256:9355b51b332c1dc18c6bd8f4968397fe173a0b001d04f1f44db4d6de67d17fad
+oid sha256:1cb0137096355587d63a085755108f414cde3b2158bccc5b9c228886410dec29
 size 29435067
diff --git a/flops-based/gpt2/13/3.pt b/flops-based/gpt2/13/3.pt
index d7b4e62017707f2f2241bfd4044b90ac60998772..4ba0baceec20466b5a36b61946d14b2866672aeb 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 sha256:2dbe21ebc29fa01c783c935e9ed61f2ec0a74e5f79827226a562689added356b
+oid sha256:c623c294d8951f605f9cb230741686c38c587edc0bcba1c86c37800f98a242b2
 size 29435067
diff --git a/flops-based/gpt2/13/4.pt b/flops-based/gpt2/13/4.pt
index c73e760202ee5ffdde6daa981620fbda446a4710..3880e7fe66f68c1a8d797fd5dcc64851aa0c0b26 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 sha256:06f2a998015ee35f53344c68af74457d2e336c9e0f3797548ee770ab27ff3eff
+oid sha256:e61eeed3404b88dcc52c6ca9d891046f8ce0b2a85f32d2893bc767d2978cb46f
 size 29435067
diff --git a/flops-based/gpt2/13/5.pt b/flops-based/gpt2/13/5.pt
index 705dd9720c524e335642acdbdb592c7301955b6a..a759b97eab431d5119ed91aba0250b0a3fa2ce85 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:c255cb2f11602524b695cf10b2b86d48a8a2527928deee874da6cea8d3bdd47a
+oid sha256:5217e17d355cfa1039fd6e84efef6497e871e299736e9c549400e052672902b0
 size 29435067
diff --git a/flops-based/gpt2/13/6.pt b/flops-based/gpt2/13/6.pt
index 3e23af637abb7b51cb2be57c1545f01159585cf3..7a53dddb5ce825746d0e9960d6f220d206eba0f1 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:c15252f3930ad4832545742db8e209f9c684cdd50c9ce738e7a4167a42290a12
+oid sha256:35323a3de8d97a82f1e1763c2ec18cf468525988449a9b4e103656d11a463a18
 size 29435067
diff --git a/flops-based/gpt2/13/7.pt b/flops-based/gpt2/13/7.pt
index 965139a24d4879c898863653203019cb12171da1..bd5e18bd206c6a29cac5536e0af7e169ef1eddfb 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:e59e13e4cab18bb564884b5620ac333806f9045108104a534fb1f3f4098a2372
+oid sha256:6289f080f0c79f48700af4b2819badb6442dc2de3931b32b7c55d3500c786fac
 size 29435067
diff --git a/flops-based/gpt2/13/8.pt b/flops-based/gpt2/13/8.pt
index 7e7941ca7872aa1aab78abd198de940a92711dec..f34965f49cfed7f0b4aa7fec5aa0691a87831f01 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 sha256:9c354337ebb059f3957e80de86407bee34764ea07887201cc6df7d5c4b81e0fe
+oid sha256:32e61a5b85fe31eafcd71e2ca6767285d48cb2821ea0929657becb6d52f9800d
 size 29435067
diff --git a/flops-based/gpt2/13/9.pt b/flops-based/gpt2/13/9.pt
index 9da7ea95a5059d4621372776fb2960d2a12582e1..2e313faa243b4f047699a4c8f07fce871985d44f 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 sha256:8d35d56a58abae15f6aedaf06086ba9cfb4fe5d7829e0c03cf95722f8ae36f89
+oid sha256:96b508a6e508fa7c6fb43adf2f19a9a26e8e30c43c71ee030190f28c7311ea66
 size 29435067
diff --git a/flops-based/gpt2/13/info.json b/flops-based/gpt2/13/info.json
index 81b0bd0fa6c8dd60aa162da7cf48220c76aaf5a7..d728c2de08968a4f19e4fae717265237bee5a362 100644
--- a/flops-based/gpt2/13/info.json
+++ b/flops-based/gpt2/13/info.json
@@ -2,71 +2,71 @@
     "shapes": [
         [
             1,
-            13
+            0
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             50257
         ]
     ],
@@ -86,19 +86,19 @@
         501765888
     ],
     "times": [
-        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
+        0.0057926177978515625,
+        0.012603759765625,
+        0.0038177967071533203,
+        0.004624128341674805,
+        0.0046122074127197266,
+        0.004325151443481445,
+        0.003463268280029297,
+        0.004107236862182617,
+        0.0040130615234375,
+        0.0020825862884521484,
+        0.0036804676055908203,
+        0.004844188690185547,
+        0.004692792892456055
     ],
     "params": [
         46471680,
diff --git a/flops-based/gpt2/14/0.pt b/flops-based/gpt2/14/0.pt
index ca2e85fb8dab81c6c710e8d77adbc1a3595f71dc..c09b16de25da3b91ac312319ba528d8e0201239d 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 sha256:043e2cf4de2b5a51303a3181fe131a83a4a7271eaa3e16b85858bc2e47338fe5
-size 186978287
+oid sha256:1061bf0650a78a55512972c3fe8862f0b8963642fec7c63eb75d0d70c4e57f05
+size 186978351
diff --git a/flops-based/gpt2/14/1.pt b/flops-based/gpt2/14/1.pt
index 21ed3e73d032e06c3a6bfc8dc2507ee93c10509d..94db289568ecc825b18d5b936b6beda923780271 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 sha256:2574f4a149ee45730de4a3e4feec5004891699d3a4bcbfaa41698b50d5d92015
+oid sha256:101f50030f1f6ab8b868a792b85cd5d6d0bf11a0b68b6d76ee7ac0bc1ddf49cb
 size 29435067
diff --git a/flops-based/gpt2/14/10.pt b/flops-based/gpt2/14/10.pt
index 9b6a33560092ea2b3664ca7ddf4bd9c22b882f4b..4f590a98555b41e3e68c2d56f1622f5145275d69 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 sha256:43828ae5893f72d2106af90e03559361f2b37fbd3c4caf2249de6e4f71760aa3
-size 29435254
+oid sha256:e37cc778eff032a4f1db57822c7182b9b4cc3ac0648cc7ce4fa57783469f2b4d
+size 29435190
diff --git a/flops-based/gpt2/14/11.pt b/flops-based/gpt2/14/11.pt
index d12aa48a9682899bea622ab1e82312c080f4945b..a1380dda85920fad85b3c4f4804ecf9d4c5abe52 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:a25cb5a6058c822d14cdd0a9643801530a409508a9b7cf9062e4663d08fce901
+oid sha256:41b4726b018ecc34310759e1a37ce6663fe38533c9cd5d7ebb6b469ad0b85a3f
 size 29435190
diff --git a/flops-based/gpt2/14/12.pt b/flops-based/gpt2/14/12.pt
index 439d29ea6d12e23869498490c93df2700bc17c3b..c0c622373521f3be70cc9c6ec8efa52355478c83 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 sha256:32bfd7809e44cd821af84faa25c9f0d361e27b6efc74b75c051ee39789240597
+oid sha256:45420eb37a00f3228f69e4eb0c8a28f92bbb2ac55c19957b36a3803cf5ad6e25
 size 12862
diff --git a/flops-based/gpt2/14/13.pt b/flops-based/gpt2/14/13.pt
index 48649f592e7da842db4222ebbfbb4f9e27bfe978..7263aec5ef95b79a27310a7c5fddfdd6f37d1b68 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 sha256:17c609defec2f68eb4193d8aa5b887972069a1a05de9093d77218b76798d7bdd
+oid sha256:80358fda99b886d845794e77b830f50964fff7e52ba348715f498fe34d723bfc
 size 154395961
diff --git a/flops-based/gpt2/14/2.pt b/flops-based/gpt2/14/2.pt
index e7646b36199ab170c84d9e72fe9b86cfdd5dabc3..b11397040a2c1c264d44950afeb1f52e64329ebd 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:dbf9a759ed80b4d663411322699c6aa97637497c9d0d3c0fe239309e10ad29c3
+oid sha256:3d17213ef617d1b3145864851847cb6f1505ab3b4144f7302b1f7323c56af54a
 size 29435067
diff --git a/flops-based/gpt2/14/3.pt b/flops-based/gpt2/14/3.pt
index 09f8dd489f3e708e51f7f6d99d729ae53588c541..bf3e46abd856910330df1f86760db219abb452cd 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 sha256:0cc687451dd6a98659f7f6f7bf81abe7967de5910a813c4d2c48f6afb6986837
+oid sha256:e32b5a67d8e1702c4ef9c0721c77afb46745bc66b39fd838114cd94474d43a65
 size 29435067
diff --git a/flops-based/gpt2/14/4.pt b/flops-based/gpt2/14/4.pt
index c8968306f06e5e8c92f6c05a8beb011864e52beb..710b424a3ca7cd86214b31e01dbbd6606105ea0f 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 sha256:2ef796a625c1f957a35f1a5916e14beced5b4877dd4c6d0178b9c193a9d649a7
+oid sha256:495f94dc08548817cf50fff7e6924352f34aa9708ca280d7b0e4ff3875ee3be3
 size 29435067
diff --git a/flops-based/gpt2/14/5.pt b/flops-based/gpt2/14/5.pt
index 1f5d1f4bdfeceae67831c01b0211fb804c33459e..4f302e356c4f8caa3234d731667cbdb71b42d65b 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 sha256:8e9c38d3f531e6cffe03aded060580a35b0a25edc02c8a0d51f35273d91c143f
+oid sha256:e1a370744ef2b3c2956325a9926bc997f7bd8fc48f509e77085bc2cde2fdd4bf
 size 29435067
diff --git a/flops-based/gpt2/14/6.pt b/flops-based/gpt2/14/6.pt
index 5df4388873a1043d319dda3e388a7adb4512146f..aad365aec8da1f492b2be075d2bdbabe26e57a54 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 sha256:2792b7f5b4b174022ea544a2c7ce1a56363ee8dd1df606a9a0eaebd6f8cb383f
+oid sha256:a91598d4e21c47aa4b3295ac1a67085a9ee1919038a9d71d1f8c4d7522424524
 size 29435067
diff --git a/flops-based/gpt2/14/7.pt b/flops-based/gpt2/14/7.pt
index 17a65d042ae483673209a5505d5dac922cde8c78..cb40634389239fd874e14e7f33b3942728868eca 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 sha256:7bba26771cd7fc604380bba9a8860c4503c868c1df984c76654b608930c57a54
+oid sha256:96f83c182f7e0fa2a595178ed561bbbede918ade289c8b1b22471447d1f80ba9
 size 29435067
diff --git a/flops-based/gpt2/14/8.pt b/flops-based/gpt2/14/8.pt
index 26ca707f398899650448e65665fd8e89376e6e33..d8faf24b47a85f436fd8c42155a264f7ddb4e3d3 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 sha256:0d2707b332eba824f423c9ba962af33d4af5705cfce5970dfc309630a6a990e6
+oid sha256:469fce7298642e9bd13215adaa6c7742f398c533ac2695c819d6b79340a88dd5
 size 29435067
diff --git a/flops-based/gpt2/14/9.pt b/flops-based/gpt2/14/9.pt
index 6771e301534aa372b469342e77d7d4d47114816c..8848eb6f1b9131247d509a24627db118f4d85f91 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 sha256:838d731be41e36d6ce344996a55023fc1abc4d4dc934cb0ae5aecb3af2feee41
+oid sha256:ba0c6c564e1ebf7d475d8a6b4305c747c7143d9823479da0b63229ede9a60527
 size 29435067
diff --git a/flops-based/gpt2/14/info.json b/flops-based/gpt2/14/info.json
index 8cbd56a8330f6dff48cd69c39f00b3a1a1298ad0..03e2e0afda3b1aee6ee9a0e823f97e6a446db198 100644
--- a/flops-based/gpt2/14/info.json
+++ b/flops-based/gpt2/14/info.json
@@ -2,76 +2,76 @@
     "shapes": [
         [
             1,
-            13
+            0
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             50257
         ]
     ],
@@ -92,20 +92,20 @@
         501765888
     ],
     "times": [
-        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
+        0.009018182754516602,
+        0.0032470226287841797,
+        0.002666950225830078,
+        0.004340648651123047,
+        0.004440784454345703,
+        0.019390344619750977,
+        0.0040683746337890625,
+        0.0021238327026367188,
+        0.002913236618041992,
+        0.002443552017211914,
+        0.002220630645751953,
+        0.0018863677978515625,
+        7.62939453125e-05,
+        0.0049130916595458984
     ],
     "params": [
         46471680,
diff --git a/flops-based/gpt2/2/0.pt b/flops-based/gpt2/2/0.pt
index 48bb3324c96ba6a435f72acd21a4a95792aa3416..b613d772eb025acc7399b8a36dec57d264d9781a 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:cfcc5c9750d6808ec396d934e483603055ce291141248477313bbcab9c217da7
-size 422422747
+oid sha256:499e886c418f7729b892d5634cb4ce6c0cb20315c5f952bd0fd9b0622114d4c3
+size 422422337
diff --git a/flops-based/gpt2/2/1.pt b/flops-based/gpt2/2/1.pt
index 30b6d2c1b67a3f28439cb7a95424ed619690f8e6..06788ef11d321ac29185a115052490e0ebeaffa4 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:bd224a80e74b2d920944bcc18a513e6e498fa79bd120e4e3e2de52a19b1d3103
-size 242697738
+oid sha256:ea24a71796233ee019e8bad4cc7cac1ebc44f03a64d79e78d8425fe01314e657
+size 242697572
diff --git a/flops-based/gpt2/2/info.json b/flops-based/gpt2/2/info.json
index 739f7f67279edc0ef82e9bb2faa5aba5e243d175..bf75790e6be6b45cc36ee8a25fb288742ea663c1 100644
--- a/flops-based/gpt2/2/info.json
+++ b/flops-based/gpt2/2/info.json
@@ -2,16 +2,16 @@
     "shapes": [
         [
             1,
-            13
+            0
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             50257
         ]
     ],
@@ -20,8 +20,8 @@
         778152960
     ],
     "times": [
-        0.014228105545043945,
-        0.012090444564819336
+        0.01551508903503418,
+        0.010844945907592773
     ],
     "params": [
         103174656,
diff --git a/flops-based/gpt2/3/0.pt b/flops-based/gpt2/3/0.pt
index d764976f6983adc599e410ff395fc0f05f965675..e0499f8b3a4faa97f6ad4b48aa6508f5bd565f15 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 sha256:4feaa64327d17762ad0a8fad9f05d4decbbb6a3501fcab5a5f9a3256045336ed
-size 334130865
+oid sha256:e9b24e59864fb9b5140868634706cdb9ddf262f5e3005566a32734962c1d414c
+size 334130545
diff --git a/flops-based/gpt2/3/1.pt b/flops-based/gpt2/3/1.pt
index e4c12f1fbddf4166267eb47978f0cd2eeb6e9df6..38820b26f7c34729cd3b1a29914bcc4268a62e08 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:fbb33507b1ed7d1d96e4d53f24641090041f1ae20f23e5b2764993cc478412f9
-size 176587645
+oid sha256:c9c845b9a6b067e4b53d1eb233a692f0842012ebccffd958f7e53c6ac2ab7d34
+size 176587397
diff --git a/flops-based/gpt2/3/2.pt b/flops-based/gpt2/3/2.pt
index 6f794fda6e386178d63a199fed7b214db5ee759b..1eae5381f4110709eca97fedcba3b513fad0ad12 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:a7c6b29459ca33d878f5f9bd34eca10b2b4e20b34620fdfb585df030f292691b
-size 154405335
+oid sha256:681ef8e1f43c806ac6c1e43769e31da715aaba5dc66a68e8d1d11cfe0ed48064
+size 154405323
diff --git a/flops-based/gpt2/3/info.json b/flops-based/gpt2/3/info.json
index 5c2a6b1b7a825c2364c92330b1b73603903ea855..760abe5b3c20317fa27c96e659018e7e5cb6ad53 100644
--- a/flops-based/gpt2/3/info.json
+++ b/flops-based/gpt2/3/info.json
@@ -2,21 +2,21 @@
     "shapes": [
         [
             1,
-            13
+            0
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             50257
         ]
     ],
@@ -26,9 +26,9 @@
         501815808
     ],
     "times": [
-        0.009680032730102539,
-        0.011411428451538086,
-        0.005023956298828125
+        0.01295781135559082,
+        0.009665727615356445,
+        0.0047380924224853516
     ],
     "params": [
         81911040,
diff --git a/flops-based/gpt2/4/0.pt b/flops-based/gpt2/4/0.pt
index 9fe3be792075aa83f10cd4d6660a251fbc321f9a..259112dcf5a87aaed9b7cc29baa175cb7521713e 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 sha256:0300334c3bf0b8a87623a0d781f4349b156d9692011b950286b02974238bdd00
-size 275269629
+oid sha256:73e8168d7f973dc68faea29c980f8a05d8b12117dfa93993b6292956c38dc931
+size 275269433
diff --git a/flops-based/gpt2/4/1.pt b/flops-based/gpt2/4/1.pt
index dbcb55e1f8e5e9fe6d18ed8dbd9c19983ba8c7fa..1932f83e6a759cae345aaa4e8cc0fae7d5b2cfcd 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 sha256:10c11a840decb6af4a241acff017cebbad7bf6158651ee78047e43d5bb281c22
-size 117726409
+oid sha256:d39310799b4aa59f623a8c414ade6f042df5596ebbc0f73378d8dd5ae2e88aff
+size 117726221
diff --git a/flops-based/gpt2/4/2.pt b/flops-based/gpt2/4/2.pt
index a95236eac225db02207ad231452fd48d29f390a9..ee8fcd19b06579d8ea82090712a1cda84f2816fa 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 sha256:010f6ec82e16f75acf74e5ab57fa8a8460df0a5e3979e031227b48b342a19c25
-size 117735929
+oid sha256:bdb92a4ff3cdbc9d8060b4ab99ebacae6680ffcca232503b8d84a1af7cee2d91
+size 117735673
diff --git a/flops-based/gpt2/4/3.pt b/flops-based/gpt2/4/3.pt
index 4d75c9b2402b6f7e2cee92cb3a161e72d3a31932..0527e67771f34ef79f95245995a29ce17f819eab 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:e645795e39feb6af486f184663727246992bf27a189c2ca8ac7ede2d40dd474a
-size 154395945
+oid sha256:382df7faa21fcbeec0563b342e48dc00491092f27b26581c571f5fdfa95d8003
+size 154395937
diff --git a/flops-based/gpt2/4/info.json b/flops-based/gpt2/4/info.json
index c456cc087940b08326f787eb9fb17426d6d5beb9..4a7a1252c8649375fa990c15bd0a6f8a45ce7acd 100644
--- a/flops-based/gpt2/4/info.json
+++ b/flops-based/gpt2/4/info.json
@@ -2,26 +2,26 @@
     "shapes": [
         [
             1,
-            13
+            0
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             50257
         ]
     ],
@@ -32,10 +32,10 @@
         501765888
     ],
     "times": [
-        0.005810737609863281,
-        0.007517337799072266,
-        0.006907463073730469,
-        0.0046427249908447266
+        0.005995750427246094,
+        0.005931854248046875,
+        0.006987094879150391,
+        0.005390644073486328
     ],
     "params": [
         67735296,
diff --git a/flops-based/gpt2/5/0.pt b/flops-based/gpt2/5/0.pt
index 7a8e89d7ba2e05eaefbecbf8bb603b61614e5fb0..0efc520cc54fa05af199a9807bd6bbdf82212e4c 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 sha256:6674582c446f71d424b7eed505dbc417967229f13dab026d7153b39acb86e64f
-size 245839011
+oid sha256:e1eabf39f2d2986108786582e7f8d2678ca3f5cb4e55f3d74e40d91f2b9655eb
+size 245838909
diff --git a/flops-based/gpt2/5/1.pt b/flops-based/gpt2/5/1.pt
index 25fb3cbe3157b783b3051692bb16268b51a6981a..5c48c4de5d015a1ad51ad63632ad78b7a183c8d3 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:d572548e583454cd22b3024d4c7d546a1c30d6cfa69d22d7eb617d0e8e26a247
-size 88295855
+oid sha256:563e8b1c4a85e683f4654b2800f2ae0fce4ac7df7988254dd283cb1bba2d73c3
+size 88295697
diff --git a/flops-based/gpt2/5/2.pt b/flops-based/gpt2/5/2.pt
index 3cbdf2d4d20932bb110ed140880ca47dcd06eb2a..ea45da31b3c88f7bf7a0d1717440300828f13885 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 sha256:042a199c02a59276e3a99aa1adc79ea62556ab051dc8166a350fc7c896184fab
-size 88295855
+oid sha256:3fdfa5573f4aebd40e45d6707b012f3da9e66930526b1b1c79ed800486a4ffdb
+size 88295697
diff --git a/flops-based/gpt2/5/3.pt b/flops-based/gpt2/5/3.pt
index f5d8c2bc46268a52b5f2a61d833ad77a26129c45..4438be6477cdc4f673f583eb40b1901ef761ffa8 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 sha256:739eeb59ee252de5302b387499478a91fca489d19de1695785f943ab08bb2b11
-size 88305311
+oid sha256:f8c4ada4096641916a374ace1e7c3d00531a6800d476c4044f9ea2cd9a09dee4
+size 88305085
diff --git a/flops-based/gpt2/5/4.pt b/flops-based/gpt2/5/4.pt
index 9bc1b620967720c0411ec7970af4732aef138f98..78c9beb2aa624c48a2e1d24cade361d49790dbb2 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:aefc14f5b41d7c02266f3cb00e51172e374af886bbdbdf6a5324d455e07957e4
-size 154395945
+oid sha256:e91513c4be51f0d6afc618504961c05ad511aea453a44a33c66dad46171b8adb
+size 154395937
diff --git a/flops-based/gpt2/5/info.json b/flops-based/gpt2/5/info.json
index 202bafde293e4a2fd4e3263e698aa98f95922ddf..0885bbe508e5fb1189b3a192b5dcda60bff08ad0 100644
--- a/flops-based/gpt2/5/info.json
+++ b/flops-based/gpt2/5/info.json
@@ -2,31 +2,31 @@
     "shapes": [
         [
             1,
-            13
+            0
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             50257
         ]
     ],
@@ -38,11 +38,11 @@
         501765888
     ],
     "times": [
-        0.005208730697631836,
-        0.005757808685302734,
-        0.00614166259765625,
-        0.005856513977050781,
-        0.004487514495849609
+        0.006539821624755859,
+        0.005666017532348633,
+        0.006500244140625,
+        0.005806684494018555,
+        0.005162477493286133
     ],
     "params": [
         60647424,
diff --git a/flops-based/gpt2/6/0.pt b/flops-based/gpt2/6/0.pt
index ba5a9950ee87a2352980ec34aab34ea6c060e9f0..cf49fb52ac8a89209832f911fd219fc22fcbd8fa 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 sha256:841ff70216a56af0e875738c5bb27e3f1be5bfaee0fa9427cfec32635bdaddf7
-size 216408649
+oid sha256:a59f3015e6b60c981ebb6d53ea5b1411857caa278eac56312be9a14fe35449f5
+size 216408513
diff --git a/flops-based/gpt2/6/1.pt b/flops-based/gpt2/6/1.pt
index 59b4a1bec2d0a4a30d0499d4f2e4b9712af76707..147faf80fed45f07949392a78fa7b24036adf833 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:c17df824a3eb5311581f7e101fcc1739c0e117058e8cf1f0299725979520f2b3
-size 58865493
+oid sha256:672587871efbf85e008af0e8630df2e9c1c0884320a287e8df052bc32a8c29ca
+size 58865365
diff --git a/flops-based/gpt2/6/2.pt b/flops-based/gpt2/6/2.pt
index d20ba9c1aaee40c785d91c65c8ff81ca12c07d53..8177d9b6612d643ff84d5850264f8acd9d163dde 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 sha256:75b5c94ced91871013df89b593311ea0a52d513c0b6026414071c269add67cd2
-size 88295855
+oid sha256:002e90e3886261dc3e2e4b54af823ef97f5ced5d6029f95a01899828b0b63406
+size 88295697
diff --git a/flops-based/gpt2/6/3.pt b/flops-based/gpt2/6/3.pt
index 334e6c98843872de92363bb9cb03193297f2555a..65aa3d41aebb82f917ccdce8f71822090ccd839f 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 sha256:6fafefa2401c77ac7e6df0da73c8d2122ba0f49eb873d120799ed31310191366
-size 88295855
+oid sha256:8428d3a810fbc26f52a0593cb9f499cb3558ac97f9c19bad3813140ee229f1c0
+size 88295697
diff --git a/flops-based/gpt2/6/4.pt b/flops-based/gpt2/6/4.pt
index c0ec2f09417fa852e0ea6c476c6ed5413813e139..af5bda368689503069386742baaf97b590e963a5 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:e928aa60585e69b12c7b427559fd50b928f4c025a56370750cfaf8261bf1f85a
-size 58874885
+oid sha256:bb3513887180cf7ec1bf5c75c44e997c1aca43161958c8150b7aa35957360fcd
+size 58874753
diff --git a/flops-based/gpt2/6/5.pt b/flops-based/gpt2/6/5.pt
index 2d8f388e885217a66f7e47a7a8ebd81e0d8bc45f..98dbab896ab561896693b5015772772c91964fee 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:d805a6074437093714986e66106975eab5d40a7b78c7c06734149d5f59c99e50
-size 154395945
+oid sha256:684f6a06eba60ae3b558a6f1a19de04ac0127dd9fd66d8770b8ef3089b805215
+size 154395937
diff --git a/flops-based/gpt2/6/info.json b/flops-based/gpt2/6/info.json
index b69aa3bad1a3091b4e257e4cfd7eec8de2df64f3..d743b0cf7d92871cf5056f4c53fbf3c916285e9f 100644
--- a/flops-based/gpt2/6/info.json
+++ b/flops-based/gpt2/6/info.json
@@ -2,36 +2,36 @@
     "shapes": [
         [
             1,
-            13
+            0
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             50257
         ]
     ],
@@ -44,12 +44,12 @@
         501765888
     ],
     "times": [
-        0.0038237571716308594,
-        0.004636526107788086,
-        0.0045757293701171875,
-        0.004657268524169922,
-        0.003826141357421875,
-        0.004747867584228516
+        0.005513668060302734,
+        0.004003047943115234,
+        0.0046880245208740234,
+        0.004880666732788086,
+        0.004079580307006836,
+        0.004527568817138672
     ],
     "params": [
         53559552,
diff --git a/flops-based/gpt2/7/0.pt b/flops-based/gpt2/7/0.pt
index bf12cbf9cd80e95ad5b17826df0b3c4dadc52803..2a59b2ad79ad4579ec8cb3683bbeee1e9e1f3b7b 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:a1248ce13b5d1c9a7d620c13518351364fed1fd49f6fd73b369a5daf8825664f
-size 216408649
+oid sha256:d886dae12b44f77a6dc78ddb0c633d061782f18cd56b6ce33226f3ca22ac34c6
+size 216408513
diff --git a/flops-based/gpt2/7/1.pt b/flops-based/gpt2/7/1.pt
index 86d9b15ebbaa6765b6db2ff18162550a3f3670f4..ffc56c03342132f914d4b12105d25ec1422b4ae5 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:d41465b818b8f34a3dddd62774013ec7c369e3e82c89d3b2e8b95bf8e8746273
-size 58865493
+oid sha256:4daeab2cfd5e175c92686bd7174bc18b420452c81e4c800c1c7c2faed645483f
+size 58865365
diff --git a/flops-based/gpt2/7/2.pt b/flops-based/gpt2/7/2.pt
index bb34208ab84af8d2e85ded4e7f451bdeaf87b2dc..d7a2b6a7bab6579bc49150b7c03691491347b6e4 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 sha256:2d3106cb68c2610d12c795e334c4a9c1ea0290547a9d0bf64b9cdcc6dc562c43
-size 58865493
+oid sha256:7959363d36a3aba919acb22f958ef17eea2eb27ed14a5b61b057f303ff15e1f8
+size 58865365
diff --git a/flops-based/gpt2/7/3.pt b/flops-based/gpt2/7/3.pt
index 9728f14c9d6d44edccd28883dd02a54108dcc491..1b5d954bf8d401d57afd51c2818db99112b93baa 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 sha256:620cf6d5bbcc468caab3b367f4db9b6045241c3ff37339423deb5f9838f4dfcc
-size 58865493
+oid sha256:cbcb22e9e43a03fa7a993dfa5d9ab4c39e5d18a231dfa8426f668ddc48e43470
+size 58865365
diff --git a/flops-based/gpt2/7/4.pt b/flops-based/gpt2/7/4.pt
index 5921387a4929384d9fcb06f854e4c3332cc43f8c..16ddb2f5fd037c3f9c80fe0bff45a3ccbef574c7 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:b21ae7fd166d4b87c6e9ac15a27744aba45e29723359599134192db25f5d321d
-size 58865493
+oid sha256:46242514cbbcacf8c05c17b6a5877422388847c9c7e29ff266deaee681231e24
+size 58865365
diff --git a/flops-based/gpt2/7/5.pt b/flops-based/gpt2/7/5.pt
index 9c67d9a5b9c755db20d6260a35b96a0d7faacec5..cbf9f9efad55328f7673b2b3b1d056748b41817a 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:d95c3e93b5c9818c5b873bec1e3447f79476e2e8685250c3c5e297db4a425449
-size 58874885
+oid sha256:4754969177e77a6e1c7bedc5637b69a750f6b78bb7f5a4d7e093ec1e65c1f5bc
+size 58874753
diff --git a/flops-based/gpt2/7/6.pt b/flops-based/gpt2/7/6.pt
index 598efd0ad7d94c363b6787edf531e3e4224f22c2..043d2a7ec3be26b408f5290a5dd9a0b6843ad154 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:c2e446a35b708b98575f36884bcbe3884047ba92df63ff361b39e6289b6c4892
-size 154395945
+oid sha256:6846f6fb00455e4150065cbd602d595bf3ca0b773cd441cf2e7caaf4361da9d3
+size 154395937
diff --git a/flops-based/gpt2/7/info.json b/flops-based/gpt2/7/info.json
index ad395956c59b4f1787ebbfb13a3c928e65c08dea..af7108108e9783e75df24cb0116463f0d2ec8754 100644
--- a/flops-based/gpt2/7/info.json
+++ b/flops-based/gpt2/7/info.json
@@ -2,41 +2,41 @@
     "shapes": [
         [
             1,
-            13
+            0
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             50257
         ]
     ],
@@ -50,13 +50,13 @@
         501765888
     ],
     "times": [
-        0.0034155845642089844,
-        0.0046579837799072266,
-        0.003651142120361328,
-        0.003980159759521484,
-        0.004025697708129883,
-        0.005561351776123047,
-        0.0044744014739990234
+        0.0035200119018554688,
+        0.003047466278076172,
+        0.015900373458862305,
+        0.004513978958129883,
+        0.00496673583984375,
+        0.0039310455322265625,
+        0.004898786544799805
     ],
     "params": [
         53559552,
diff --git a/flops-based/gpt2/8/0.pt b/flops-based/gpt2/8/0.pt
index 7b93200556da84365eea9924d64bf4ed93d27b5d..99c8b79acc279a488db52dc9d64b3f6a0fcad33f 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:b25cbd1372d6775a4d2015bdc457f74a79a74c43027d37164004898d5821dbfb
-size 186978287
+oid sha256:b6e498c0f3deda34e28bf982c4440bbb1b5d4377d0f310f5f3839332e270983b
+size 186978245
diff --git a/flops-based/gpt2/8/1.pt b/flops-based/gpt2/8/1.pt
index d04df8f2f605fd26998b032bc8a242fb8b71da01..a64a7d0adeec1a398f2fb3aceade9584cbc270b7 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 sha256:102d2f0d7faed9e422bc1e3d931d44fdf57f29e4f139f117effcabf00b7dfca3
-size 58865493
+oid sha256:f69f67d6c7695ce16ef95750e5d1d9c0995140c7746589e70247739f40aae864
+size 58865365
diff --git a/flops-based/gpt2/8/2.pt b/flops-based/gpt2/8/2.pt
index 204599291a1e43a1ffec3ea9ca853ffd955dfd9f..496e5bf64c4ed7078062c054ea74027fc024d3e0 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 sha256:3e024cfb3fafa62bb728b1842319d3954ef6d4001c3d488d9ad0f57e77eabc27
-size 58865493
+oid sha256:e368f9340ffddcc57702af81fc04bb350e709eead50351a7c464a5ab9fa32be6
+size 58865365
diff --git a/flops-based/gpt2/8/3.pt b/flops-based/gpt2/8/3.pt
index 9af5055f90f62f909e074f336f8ebf72871cd86f..05732b2930293911273990fcbd0c9652028dcf3d 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 sha256:45f0ed0cb0a262b0a68b76711d78cd5c43676f95ca041cbed8775ea3e47bebe3
-size 58865493
+oid sha256:3927892b41e685991192a3e5b67b8f49c2348799969b8e3e259a81c99f0f06e1
+size 58865365
diff --git a/flops-based/gpt2/8/4.pt b/flops-based/gpt2/8/4.pt
index 12c929d3c8b3dffd5107d9b79ac80b73e79a6b0b..e1474a81a3f81a5e669bebc1e90a015c2f587562 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 sha256:5fc20eb876af3aba16897aca4f1960747187ecc37ff67c986c406f9497ecee29
-size 58865493
+oid sha256:54c062b11e2edbcb88aadaf60fe37255df09e878bcbc5416056d3072cde6d3fc
+size 58865365
diff --git a/flops-based/gpt2/8/5.pt b/flops-based/gpt2/8/5.pt
index c973018c73673ee86de412cc632a65aae7714bd8..a5a6045e5b056a4327606fd3182d5f098e57b36f 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 sha256:37011e66feb3ef65acbe6228b95f61e0508022dc06b7db1608c69ce220489bd5
-size 58865493
+oid sha256:de0c418fb11c383d2c58e3cb11d706b43403b8befacf0825dc14c18c30fc86fc
+size 58865365
diff --git a/flops-based/gpt2/8/6.pt b/flops-based/gpt2/8/6.pt
index 73f30873ecebde6b4efebef5e4f59526ad37170c..20890ea2762f3ca2be42507dfdb068f00c19aeb2 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:dddff8a48436789d1ce7ef922c3b0fb9b3126ffd692c198038f39420d732baab
-size 29444523
+oid sha256:f1e021188d27e50ecfad815903a8bb78691a6faae46e52b411a901946f64651d
+size 29444485
diff --git a/flops-based/gpt2/8/7.pt b/flops-based/gpt2/8/7.pt
index 525b27506aadc90d578377986f5499b9b2504fb4..130e29a2c1f5788186732b548bfccd03f934dded 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 sha256:5dc01f4dd309849cb2a7fbf367b1db74b02418adc0f52f1cd877b04f4336bb40
-size 154395945
+oid sha256:adf1bc5ddf1cd4a171ac44acc2172d61a827522310db798128048ea8b9cc6358
+size 154395937
diff --git a/flops-based/gpt2/8/info.json b/flops-based/gpt2/8/info.json
index 5d34c1bdeb03d5c347d5d85e5d7a48601b409c30..e7d01cb1d04cb6f9f9dd61067d61ee6135898544 100644
--- a/flops-based/gpt2/8/info.json
+++ b/flops-based/gpt2/8/info.json
@@ -2,46 +2,46 @@
     "shapes": [
         [
             1,
-            13
+            0
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             50257
         ]
     ],
@@ -56,14 +56,14 @@
         501765888
     ],
     "times": [
-        0.0025146007537841797,
-        0.004233360290527344,
-        0.0049517154693603516,
-        0.0041806697845458984,
-        0.003915309906005859,
-        0.0038356781005859375,
-        0.0019066333770751953,
-        0.004812002182006836
+        0.001880645751953125,
+        0.0029189586639404297,
+        0.003826141357421875,
+        0.0045244693756103516,
+        0.0038902759552001953,
+        0.0036885738372802734,
+        0.0017671585083007812,
+        0.009768486022949219
     ],
     "params": [
         46471680,
diff --git a/flops-based/gpt2/9/0.pt b/flops-based/gpt2/9/0.pt
index 8fe8fb1b6ef4a11047e9a995921d9955bf54fc51..4f75f891f77f3779e26199fbec990cd72b05df04 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 sha256:3e6ab798cdf89bd919b41534fc30a8db4cf57d60a3b9c405ff13d26c2f29af19
-size 186978287
+oid sha256:7c9dc4f46354c72dfc3f9fca1b57ff361415e7809cd1d75cdd7f0bc87d43e2c6
+size 186978245
diff --git a/flops-based/gpt2/9/1.pt b/flops-based/gpt2/9/1.pt
index a6f0590def02569a9f4450645c501b8a2af74fde..598726c1cd6f3f198e11d1971f5653adbd494d35 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 sha256:04e26de9ea7dcb25f7b95b64715a9e0a7778bc3da8ff5725caa8e9ad431ddc49
-size 29435067
+oid sha256:cab6c806fca8acf129d3c868210649222153f578d23c17e9a163806caca18fa6
+size 29434969
diff --git a/flops-based/gpt2/9/2.pt b/flops-based/gpt2/9/2.pt
index 6aa7d01689e13cecad35e6f1dc60700cd7952c80..039d8d239e3906fd502b4a7cdd71390f84ce7a4c 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:b72bdffa2b72c324bbb6c3f290e776f072ba3f5c79bc21b269c56fa36c7e76c3
-size 29435067
+oid sha256:e71f41e7220afe2d7598064fcaed0fdcf8df130a25e3cf5a36fb639417905769
+size 29434905
diff --git a/flops-based/gpt2/9/3.pt b/flops-based/gpt2/9/3.pt
index c5c0a8740b8d28cbcc6b7b25145fe744f5460195..82d0aec58de9fba8764338882042c640d705e5ad 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:c6fbdc21d0b494a55545a3510d77c1984a3c21ded7644520ba146d82f2ff6852
-size 58865493
+oid sha256:11186b122d936fb9059e38b42711e60a58d21fc28728a1a5fef98d42a561f82a
+size 58865365
diff --git a/flops-based/gpt2/9/4.pt b/flops-based/gpt2/9/4.pt
index 65892717f6ec939747ba61fa8bd1f8a129a78688..9e5908517da8d42b773fdf0962ac012033a65e0f 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 sha256:3d10682b3bcd75c2e86f971b0315c6aef8e10c02103c2c2309dcb29acfca5bd6
-size 58865493
+oid sha256:e97fd1c52de37f26bc105b7b946fb15690b5000ea788d4b43c99f3969e6c34a7
+size 58865365
diff --git a/flops-based/gpt2/9/5.pt b/flops-based/gpt2/9/5.pt
index a9530bd8d75045aee83e558c151fc74da95098a1..5714f0e9e4d4e85ddaee135adfc7b0e87cace39e 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 sha256:35e208d811b3d0178137fc692f975b9cd199fc4ac8002579d4c21534fbc34ef1
-size 58865493
+oid sha256:f872640e93179381d5749ae959ba0e1cfe95822e8712c6d78ee2b280f3c8207c
+size 58865365
diff --git a/flops-based/gpt2/9/6.pt b/flops-based/gpt2/9/6.pt
index 22e59f1ec675f47757162aa84779b55de672283c..94546391152d1a23161e60433eecc5f1d3c35d77 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 sha256:0ae8c0de8a4076b5621e1df894237896b1635fe73295edc493e9d62a2c60316c
-size 58865493
+oid sha256:9531ad4489155acd480434f097f4767dcb88beefe3a02f5001e1f7dc16c713a5
+size 58865365
diff --git a/flops-based/gpt2/9/7.pt b/flops-based/gpt2/9/7.pt
index e2a2fc4e8f009cfc0f14852d61ee8444729166d9..20d4d97194160c4d1c8bb53795b8eb3afedf03a8 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:b300388569a299fa3e402a9fd6d8fb3359ca8af02fffd256756ac336a69b6827
-size 29444523
+oid sha256:684a84659a591624deef9b0589b167695201c0c505f037bcba6d867f2f846475
+size 29444421
diff --git a/flops-based/gpt2/9/8.pt b/flops-based/gpt2/9/8.pt
index a28f59e099c7c66e48a045e00f87e863254fa2ca..af6b99ef135d2e33986980050c9b93b7567e61d8 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 sha256:5285f113a6dbdcf7a84095e9424e74eb5f3919879aeb2dad7bf009c22f70c3be
-size 154395945
+oid sha256:e1e92d47025f6bdc6bad828e876a281e36080a25ad841b2295648146c9032cf3
+size 154395937
diff --git a/flops-based/gpt2/9/info.json b/flops-based/gpt2/9/info.json
index 474300bb5f5768b73235895b3bd4bba57b057259..5ab119db1c72c1a5d731dde94c3123a33eb77b4a 100644
--- a/flops-based/gpt2/9/info.json
+++ b/flops-based/gpt2/9/info.json
@@ -2,51 +2,51 @@
     "shapes": [
         [
             1,
-            13
+            0
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             768
         ],
         [
             1,
-            13,
+            0,
             50257
         ]
     ],
@@ -62,15 +62,15 @@
         501765888
     ],
     "times": [
-        0.004093170166015625,
-        0.0020084381103515625,
-        0.001920461654663086,
-        0.005430698394775391,
-        0.0036170482635498047,
-        0.004530429840087891,
-        0.00299835205078125,
-        0.0017931461334228516,
-        0.004969120025634766
+        0.0027399063110351562,
+        0.003090381622314453,
+        0.002351999282836914,
+        0.004258155822753906,
+        0.004597663879394531,
+        0.005335092544555664,
+        0.004565238952636719,
+        0.004034996032714844,
+        0.005158901214599609
     ],
     "params": [
         46471680,