Upload folder using huggingface_hub

same_total_bs_half_seq_len/args.json

@@ -0,0 +1 @@
+{"hellaswag": true, "attention_kind": "selective", "log_dir": "same_total_bs_half_seq_len", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 12, "protect_bos_token": true, "prevent_from_masking_myself": true, "max_steps": null, "warmup_steps": null, "group": "shrinking_big_runs_2", "use_wandb": true, "kill_self_after_run": false, "random_seed": 1337, "memory_penalty_epsilon": 0.1, "selection_head_linear_combo": "none", "selection_head_linear_combo_scale": 1.0, "protection_kind": "none", "leaky_relu_alpha": null, "leaky_relu_bias": null, "use_compile": true, "use_mini_model": false, "upload_to_hf": true, "seq_len": 512, "batch_size": 16, "total_batch_size": 524288, "protection_head_scaling_factor": 1.0, "protection_head_bias": 0.0, "n_sliced_masks": null, "n_latent_masks": null, "mask_layernorm": false, "residual_attention_masks": false, "compute_base_shapes": false, "base_shapes_savefile": null, "mup": false, "disable_selection": false, "mup_enable_coord_check_logging": false, "max_lr": 6e-05, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": null, "n_embd": 768}

same_total_bs_half_seq_len/dataloader_02499.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55666202dc65099e5a26b66e4e3e7470de2cc1c0921ebdf41ac539630f8a4ecd
+size 964

same_total_bs_half_seq_len/log2.txt

@@ -0,0 +1,303 @@
+max_steps: 2500
+0 val loss 10.9883
+0 val perplexity 59177.6016
+0 train 10.988361 (lr=8.3916e-08) (hash(x)=41882737)
+10 train 10.964879 (lr=9.2308e-07) (hash(x)=38752233)
+20 train 10.901049 (lr=1.7622e-06) (hash(x)=41987968)
+30 train 10.798004 (lr=2.6014e-06) (hash(x)=39173693)
+40 train 10.647665 (lr=3.4406e-06) (hash(x)=44364470)
+50 train 10.456973 (lr=4.2797e-06) (hash(x)=37915340)
+60 train 10.232116 (lr=5.1189e-06) (hash(x)=40075499)
+70 train 10.034853 (lr=5.9580e-06) (hash(x)=39209274)
+80 train 9.879135 (lr=6.7972e-06) (hash(x)=35191202)
+90 train 9.791026 (lr=7.6364e-06) (hash(x)=40296838)
+100 val loss 9.7028
+100 val perplexity 16362.9043
+100 train 9.711576 (lr=8.4755e-06) (hash(x)=38049409)
+110 train 9.636579 (lr=9.3147e-06) (hash(x)=49174314)
+120 train 9.611252 (lr=1.0154e-05) (hash(x)=39993942)
+130 train 9.553054 (lr=1.0993e-05) (hash(x)=34436824)
+140 train 9.432757 (lr=1.1832e-05) (hash(x)=35630599)
+150 train 9.366984 (lr=1.2671e-05) (hash(x)=38098135)
+160 train 9.274781 (lr=1.3510e-05) (hash(x)=46948704)
+170 train 9.166862 (lr=1.4350e-05) (hash(x)=41276821)
+180 train 9.045074 (lr=1.5189e-05) (hash(x)=37776315)
+190 train 8.941359 (lr=1.6028e-05) (hash(x)=39307311)
+200 val loss 8.7803
+200 val perplexity 6505.0151
+200 train 8.768229 (lr=1.6867e-05) (hash(x)=38965293)
+210 train 8.612945 (lr=1.7706e-05) (hash(x)=43574397)
+220 train 8.504194 (lr=1.8545e-05) (hash(x)=47662442)
+230 train 8.336093 (lr=1.9385e-05) (hash(x)=41008662)
+240 train 8.183731 (lr=2.0224e-05) (hash(x)=37003526)
+250 train 8.015138 (lr=2.1063e-05) (hash(x)=35863708)
+260 train 7.878811 (lr=2.1902e-05) (hash(x)=41433145)
+270 train 7.902848 (lr=2.2741e-05) (hash(x)=39955434)
+280 train 7.732792 (lr=2.3580e-05) (hash(x)=42053458)
+290 train 7.628213 (lr=2.4420e-05) (hash(x)=38754802)
+300 val loss 7.6190
+300 val perplexity 2036.4838
+300 train 7.502206 (lr=2.5259e-05) (hash(x)=37473643)
+310 train 7.573412 (lr=2.6098e-05) (hash(x)=41946823)
+320 train 7.496784 (lr=2.6937e-05) (hash(x)=37217925)
+330 train 7.369864 (lr=2.7776e-05) (hash(x)=48653656)
+340 train 7.279233 (lr=2.8615e-05) (hash(x)=36993091)
+350 train 7.269088 (lr=2.9455e-05) (hash(x)=38742081)
+360 train 7.358791 (lr=3.0294e-05) (hash(x)=43576901)
+370 train 7.192691 (lr=3.1133e-05) (hash(x)=36570538)
+380 train 7.158815 (lr=3.1972e-05) (hash(x)=43348240)
+390 train 7.205514 (lr=3.2811e-05) (hash(x)=37875659)
+400 val loss 7.1939
+400 val perplexity 1331.2561
+400 train 7.222857 (lr=3.3650e-05) (hash(x)=41660812)
+410 train 7.080593 (lr=3.4490e-05) (hash(x)=37968408)
+420 train 7.049312 (lr=3.5329e-05) (hash(x)=41080176)
+430 train 7.129427 (lr=3.6168e-05) (hash(x)=34708528)
+440 train 7.016343 (lr=3.7007e-05) (hash(x)=41041500)
+450 train 7.004900 (lr=3.7846e-05) (hash(x)=37590081)
+460 train 6.931201 (lr=3.8685e-05) (hash(x)=42370259)
+470 train 6.854811 (lr=3.9524e-05) (hash(x)=38174196)
+480 train 7.027260 (lr=4.0364e-05) (hash(x)=39264243)
+490 train 6.937497 (lr=4.1203e-05) (hash(x)=39565440)
+500 val loss 6.9632
+500 val perplexity 1057.0006
+500 train 6.838260 (lr=4.2042e-05) (hash(x)=35815196)
+510 train 6.785631 (lr=4.2881e-05) (hash(x)=39616059)
+520 train 6.830124 (lr=4.3720e-05) (hash(x)=49078223)
+530 train 6.999103 (lr=4.4559e-05) (hash(x)=45929312)
+540 train 6.825980 (lr=4.5399e-05) (hash(x)=39860869)
+550 train 6.768868 (lr=4.6238e-05) (hash(x)=45524200)
+560 train 6.713850 (lr=4.7077e-05) (hash(x)=34657559)
+570 train 6.870642 (lr=4.7916e-05) (hash(x)=34198027)
+580 train 6.861985 (lr=4.8755e-05) (hash(x)=44856508)
+590 train 6.711569 (lr=4.9594e-05) (hash(x)=42146659)
+600 val loss 6.7726
+600 val perplexity 873.5662
+600 train 6.638854 (lr=5.0434e-05) (hash(x)=36673381)
+610 train 6.642494 (lr=5.1273e-05) (hash(x)=46081639)
+620 train 6.620163 (lr=5.2112e-05) (hash(x)=41640874)
+630 train 6.654254 (lr=5.2951e-05) (hash(x)=41224300)
+640 train 6.619634 (lr=5.3790e-05) (hash(x)=39147021)
+650 train 6.615162 (lr=5.4629e-05) (hash(x)=36710170)
+660 train 6.514262 (lr=5.5469e-05) (hash(x)=41488466)
+670 train 6.646647 (lr=5.6308e-05) (hash(x)=41010960)
+680 train 6.653363 (lr=5.7147e-05) (hash(x)=39429901)
+690 train 6.649450 (lr=5.7986e-05) (hash(x)=44800066)
+700 val loss 6.6007
+700 val perplexity 735.5787
+700 train 6.542794 (lr=5.8825e-05) (hash(x)=46954230)
+710 train 6.348343 (lr=5.9664e-05) (hash(x)=44729399)
+720 train 6.597054 (lr=5.9999e-05) (hash(x)=40288354)
+730 train 6.446252 (lr=5.9991e-05) (hash(x)=44089962)
+740 train 6.396541 (lr=5.9974e-05) (hash(x)=40026145)
+750 train 6.365641 (lr=5.9949e-05) (hash(x)=36468229)
+760 train 6.310937 (lr=5.9915e-05) (hash(x)=39182712)
+770 train 6.324027 (lr=5.9874e-05) (hash(x)=41005834)
+780 train 6.346465 (lr=5.9824e-05) (hash(x)=38232985)
+790 train 6.503117 (lr=5.9765e-05) (hash(x)=38064072)
+800 val loss 6.4420
+800 val perplexity 627.6665
+800 train 6.359019 (lr=5.9698e-05) (hash(x)=39571816)
+810 train 6.239057 (lr=5.9623e-05) (hash(x)=41286196)
+820 train 6.412337 (lr=5.9540e-05) (hash(x)=44007678)
+830 train 6.315421 (lr=5.9449e-05) (hash(x)=36771154)
+840 train 6.240267 (lr=5.9349e-05) (hash(x)=44197362)
+850 train 6.258144 (lr=5.9241e-05) (hash(x)=37351368)
+860 train 6.343967 (lr=5.9126e-05) (hash(x)=42629178)
+870 train 6.260613 (lr=5.9002e-05) (hash(x)=36215816)
+880 train 6.183096 (lr=5.8869e-05) (hash(x)=45418370)
+890 train 6.388379 (lr=5.8729e-05) (hash(x)=38983291)
+900 val loss 6.3183
+900 val perplexity 554.6538
+900 train 6.229974 (lr=5.8581e-05) (hash(x)=34016998)
+910 train 6.147171 (lr=5.8425e-05) (hash(x)=36922232)
+920 train 6.201122 (lr=5.8262e-05) (hash(x)=43334032)
+930 train 6.264863 (lr=5.8090e-05) (hash(x)=39139652)
+940 train 6.182590 (lr=5.7911e-05) (hash(x)=39832832)
+950 train 6.129735 (lr=5.7723e-05) (hash(x)=38795848)
+960 train 6.159459 (lr=5.7529e-05) (hash(x)=37215711)
+970 train 6.121268 (lr=5.7326e-05) (hash(x)=37313575)
+980 train 6.297005 (lr=5.7116e-05) (hash(x)=46141234)
+990 train 6.203898 (lr=5.6899e-05) (hash(x)=41240889)
+1000 val loss 6.2156
+1000 val perplexity 500.5182
+1000 train 6.064166 (lr=5.6674e-05) (hash(x)=40204404)
+1010 train 6.196434 (lr=5.6442e-05) (hash(x)=34594593)
+1020 train 6.173024 (lr=5.6202e-05) (hash(x)=38795311)
+1030 train 6.064879 (lr=5.5956e-05) (hash(x)=43467681)
+1040 train 6.061503 (lr=5.5702e-05) (hash(x)=38728767)
+1050 train 6.174935 (lr=5.5441e-05) (hash(x)=42307606)
+1060 train 6.149168 (lr=5.5174e-05) (hash(x)=43499838)
+1070 train 6.018306 (lr=5.4899e-05) (hash(x)=42853705)
+1080 train 6.131411 (lr=5.4618e-05) (hash(x)=38874363)
+1090 train 6.100400 (lr=5.4330e-05) (hash(x)=44247152)
+1100 val loss 6.1357
+1100 val perplexity 462.0478
+1100 train 6.009881 (lr=5.4035e-05) (hash(x)=41926226)
+1110 train 6.155123 (lr=5.3734e-05) (hash(x)=42645790)
+1120 train 6.127011 (lr=5.3426e-05) (hash(x)=43723936)
+1130 train 6.054988 (lr=5.3113e-05) (hash(x)=41543766)
+1140 train 5.968549 (lr=5.2792e-05) (hash(x)=39842285)
+1150 train 5.949434 (lr=5.2466e-05) (hash(x)=41160965)
+1160 train 6.095147 (lr=5.2134e-05) (hash(x)=42843998)
+1170 train 6.027193 (lr=5.1796e-05) (hash(x)=30950803)
+1180 train 6.003647 (lr=5.1452e-05) (hash(x)=34820484)
+1190 train 5.960533 (lr=5.1102e-05) (hash(x)=37698786)
+1200 val loss 6.0610
+1200 val perplexity 428.8246
+1200 train 5.945203 (lr=5.0746e-05) (hash(x)=41861487)
+1210 train 6.122990 (lr=5.0386e-05) (hash(x)=39863630)
+1220 train 6.049147 (lr=5.0019e-05) (hash(x)=40242551)
+1230 train 5.914651 (lr=4.9648e-05) (hash(x)=44311591)
+1240 train 5.899943 (lr=4.9271e-05) (hash(x)=39781776)
+1250 train 6.057640 (lr=4.8889e-05) (hash(x)=42390421)
+1260 train 6.042126 (lr=4.8503e-05) (hash(x)=41152137)
+1270 train 5.929941 (lr=4.8111e-05) (hash(x)=36946170)
+1280 train 5.884705 (lr=4.7715e-05) (hash(x)=37347986)
+1290 train 5.868985 (lr=4.7315e-05) (hash(x)=32033927)
+1300 val loss 5.9937
+1300 val perplexity 400.9005
+1300 train 5.995255 (lr=4.6909e-05) (hash(x)=44258036)
+1310 train 5.955841 (lr=4.6500e-05) (hash(x)=46067973)
+1320 train 5.878343 (lr=4.6086e-05) (hash(x)=39193154)
+1330 train 5.865586 (lr=4.5669e-05) (hash(x)=45642219)
+1340 train 5.918052 (lr=4.5247e-05) (hash(x)=40763259)
+1350 train 5.978745 (lr=4.4822e-05) (hash(x)=38172293)
+1360 train 5.876603 (lr=4.4393e-05) (hash(x)=37114843)
+1370 train 5.899178 (lr=4.3960e-05) (hash(x)=40819318)
+1380 train 5.924349 (lr=4.3524e-05) (hash(x)=42057083)
+1390 train 5.898895 (lr=4.3085e-05) (hash(x)=38347811)
+1400 val loss 5.9387
+1400 val perplexity 379.4548
+1400 train 5.839736 (lr=4.2643e-05) (hash(x)=35847481)
+1410 train 5.912867 (lr=4.2197e-05) (hash(x)=38699246)
+1420 train 5.906290 (lr=4.1749e-05) (hash(x)=38324524)
+1430 train 5.861506 (lr=4.1298e-05) (hash(x)=40369167)
+1440 train 5.848300 (lr=4.0845e-05) (hash(x)=33500630)
+1450 train 5.871412 (lr=4.0389e-05) (hash(x)=42605844)
+1460 train 5.847542 (lr=3.9931e-05) (hash(x)=35638705)
+1470 train 5.776109 (lr=3.9470e-05) (hash(x)=44425926)
+1480 train 5.927655 (lr=3.9008e-05) (hash(x)=45980198)
+1490 train 5.874119 (lr=3.8544e-05) (hash(x)=36381167)
+1500 val loss 5.8856
+1500 val perplexity 359.8066
+1500 train 5.797331 (lr=3.8078e-05) (hash(x)=39218969)
+1510 train 5.803570 (lr=3.7610e-05) (hash(x)=34963329)
+1520 train 5.910394 (lr=3.7142e-05) (hash(x)=38493490)
+1530 train 5.809539 (lr=3.6671e-05) (hash(x)=37779234)
+1540 train 5.820983 (lr=3.6200e-05) (hash(x)=39600724)
+1550 train 5.892306 (lr=3.5728e-05) (hash(x)=41652676)
+1560 train 5.987523 (lr=3.5255e-05) (hash(x)=36794158)
+1570 train 5.780287 (lr=3.4781e-05) (hash(x)=28528384)
+1580 train 5.854288 (lr=3.4306e-05) (hash(x)=40202410)
+1590 train 5.799783 (lr=3.3831e-05) (hash(x)=41923463)
+1600 val loss 5.8424
+1600 val perplexity 344.6087
+1600 train 5.722152 (lr=3.3356e-05) (hash(x)=39934496)
+1610 train 5.763445 (lr=3.2881e-05) (hash(x)=40576932)
+1620 train 5.976724 (lr=3.2406e-05) (hash(x)=41986373)
+1630 train 5.841710 (lr=3.1931e-05) (hash(x)=38502265)
+1640 train 5.739820 (lr=3.1456e-05) (hash(x)=38956594)
+1650 train 5.813583 (lr=3.0982e-05) (hash(x)=37470941)
+1660 train 5.759623 (lr=3.0509e-05) (hash(x)=35716340)
+1670 train 5.694458 (lr=3.0036e-05) (hash(x)=39081951)
+1680 train 5.939166 (lr=2.9564e-05) (hash(x)=44126819)
+1690 train 5.783695 (lr=2.9093e-05) (hash(x)=39815554)
+1700 val loss 5.8039
+1700 val perplexity 331.5769
+1700 train 5.794112 (lr=2.8624e-05) (hash(x)=37646590)
+1710 train 5.688335 (lr=2.8156e-05) (hash(x)=38286621)
+1720 train 5.831466 (lr=2.7689e-05) (hash(x)=38421820)
+1730 train 5.753751 (lr=2.7224e-05) (hash(x)=41288435)
+1740 train 5.709715 (lr=2.6761e-05) (hash(x)=45642859)
+1750 train 5.613677 (lr=2.6299e-05) (hash(x)=36528352)
+1760 train 5.812756 (lr=2.5840e-05) (hash(x)=36914578)
+1770 train 5.726003 (lr=2.5383e-05) (hash(x)=38009157)
+1780 train 5.654666 (lr=2.4928e-05) (hash(x)=36105857)
+1790 train 5.800747 (lr=2.4476e-05) (hash(x)=40873205)
+1800 val loss 5.7709
+1800 val perplexity 320.8292
+1800 train 5.706655 (lr=2.4026e-05) (hash(x)=31431733)
+1810 train 5.680541 (lr=2.3580e-05) (hash(x)=37889200)
+1820 train 5.670816 (lr=2.3136e-05) (hash(x)=41974666)
+1830 train 5.773417 (lr=2.2695e-05) (hash(x)=38425824)
+1840 train 5.655602 (lr=2.2257e-05) (hash(x)=40983908)
+1850 train 5.683918 (lr=2.1823e-05) (hash(x)=43685380)
+1860 train 5.780799 (lr=2.1392e-05) (hash(x)=31777770)
+1870 train 5.632263 (lr=2.0965e-05) (hash(x)=44109703)
+1880 train 5.635700 (lr=2.0542e-05) (hash(x)=35678244)
+1890 train 5.711023 (lr=2.0122e-05) (hash(x)=35649940)
+1900 val loss 5.7454
+1900 val perplexity 312.7442
+1900 train 5.804367 (lr=1.9706e-05) (hash(x)=43009344)
+1910 train 5.723359 (lr=1.9295e-05) (hash(x)=41641641)
+1920 train 5.678304 (lr=1.8887e-05) (hash(x)=41378670)
+1930 train 5.647694 (lr=1.8485e-05) (hash(x)=42228302)
+1940 train 5.713704 (lr=1.8086e-05) (hash(x)=35659474)
+1950 train 5.632979 (lr=1.7692e-05) (hash(x)=43019440)
+1960 train 5.635161 (lr=1.7303e-05) (hash(x)=36388055)
+1970 train 5.731670 (lr=1.6919e-05) (hash(x)=38467899)
+1980 train 5.664096 (lr=1.6540e-05) (hash(x)=37166112)
+1990 train 5.721681 (lr=1.6166e-05) (hash(x)=39116198)
+2000 val loss 5.7242
+2000 val perplexity 306.1873
+2000 train 5.568165 (lr=1.5797e-05) (hash(x)=34741552)
+2010 train 5.747209 (lr=1.5433e-05) (hash(x)=39795810)
+2020 train 5.672577 (lr=1.5075e-05) (hash(x)=39901562)
+2030 train 5.620083 (lr=1.4723e-05) (hash(x)=32955586)
+2040 train 5.700198 (lr=1.4376e-05) (hash(x)=40879614)
+2050 train 5.691049 (lr=1.4035e-05) (hash(x)=39268533)
+2060 train 5.614498 (lr=1.3699e-05) (hash(x)=44795608)
+2070 train 5.580024 (lr=1.3370e-05) (hash(x)=39229893)
+2080 train 5.726781 (lr=1.3047e-05) (hash(x)=37103514)
+2090 train 5.649680 (lr=1.2730e-05) (hash(x)=40987512)
+2100 val loss 5.7037
+2100 val perplexity 299.9721
+2100 train 5.701947 (lr=1.2419e-05) (hash(x)=51452794)
+2110 train 5.701648 (lr=1.2115e-05) (hash(x)=42262641)
+2120 train 5.639098 (lr=1.1817e-05) (hash(x)=41984974)
+2130 train 5.586434 (lr=1.1525e-05) (hash(x)=35439208)
+2140 train 5.716301 (lr=1.1241e-05) (hash(x)=39817069)
+2150 train 5.647588 (lr=1.0963e-05) (hash(x)=37989592)
+2160 train 5.565844 (lr=1.0692e-05) (hash(x)=49885738)
+2170 train 5.748156 (lr=1.0427e-05) (hash(x)=40654241)
+2180 train 5.672402 (lr=1.0170e-05) (hash(x)=41950889)
+2190 train 5.691886 (lr=9.9199e-06) (hash(x)=40294158)
+2200 val loss 5.6887
+2200 val perplexity 295.5184
+2200 train 5.597044 (lr=9.6769e-06) (hash(x)=40076244)
+2210 train 5.645071 (lr=9.4412e-06) (hash(x)=36688120)
+2220 train 5.623955 (lr=9.2127e-06) (hash(x)=37605188)
+2230 train 5.606855 (lr=8.9916e-06) (hash(x)=35632798)
+2240 train 5.697649 (lr=8.7779e-06) (hash(x)=42825466)
+2250 train 5.646768 (lr=8.5717e-06) (hash(x)=35071109)
+2260 train 5.607198 (lr=8.3731e-06) (hash(x)=47897078)
+2270 train 5.624213 (lr=8.1821e-06) (hash(x)=46171759)
+2280 train 5.694666 (lr=7.9988e-06) (hash(x)=41213863)
+2290 train 5.606037 (lr=7.8232e-06) (hash(x)=40640886)
+2300 val loss 5.6771
+2300 val perplexity 292.0914
+2300 train 5.524591 (lr=7.6555e-06) (hash(x)=44989210)
+2310 train 5.669634 (lr=7.4956e-06) (hash(x)=40352035)
+2320 train 5.651531 (lr=7.3436e-06) (hash(x)=42788091)
+2330 train 5.552987 (lr=7.1995e-06) (hash(x)=40234639)
+2340 train 5.564157 (lr=7.0635e-06) (hash(x)=35472103)
+2350 train 5.693839 (lr=6.9354e-06) (hash(x)=42043124)
+2360 train 5.547943 (lr=6.8155e-06) (hash(x)=39972275)
+2370 train 5.592579 (lr=6.7036e-06) (hash(x)=36566627)
+2380 train 5.727581 (lr=6.5999e-06) (hash(x)=55698840)
+2390 train 5.616693 (lr=6.5044e-06) (hash(x)=41155464)
+2400 val loss 5.6679
+2400 val perplexity 289.4286
+2400 train 5.578992 (lr=6.4171e-06) (hash(x)=43485840)
+2410 train 5.555132 (lr=6.3380e-06) (hash(x)=39685043)
+2420 train 5.710906 (lr=6.2672e-06) (hash(x)=39496037)
+2430 train 5.574580 (lr=6.2046e-06) (hash(x)=41962948)
+2440 train 5.547521 (lr=6.1504e-06) (hash(x)=34986640)
+2450 train 5.662853 (lr=6.1045e-06) (hash(x)=39123972)
+2460 train 5.540375 (lr=6.0669e-06) (hash(x)=38112131)
+2470 train 5.560893 (lr=6.0376e-06) (hash(x)=39029104)
+2480 train 5.494836 (lr=6.0167e-06) (hash(x)=38048270)
+2490 train 5.635230 (lr=6.0042e-06) (hash(x)=37369977)
+2499 val loss 5.6608
+2499 val perplexity 287.3667

same_total_bs_half_seq_len/model_02499.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a2a356733bd154826b1812685e5da6e763f5377c7c97d143bff4777185295a13
+size 508831106

same_total_bs_half_seq_len/optimizer_02499.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b00727bca23ecfe8851d8961a82a974d52ef49c4895e9da052d22481b145fc69
+size 992507142