Upload folder using huggingface_hub

attention_kindselective_n_heads4_seed1341/args.json

@@ -1 +1 @@
-{"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_6/attention_kindselective_n_heads4_seed1341", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 4, "protect_bos_token": true, "prevent_from_masking_myself": true, "max_steps": 10000, "warmup_steps": 200, "group": "wider_is_better_6", "use_wandb": true, "kill_self_after_run": false, "random_seed": 1341, "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": 256, "batch_size": 40, "total_batch_size": 10240, "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": true, "disable_selection": false, "mup_enable_coord_check_logging": false, "max_lr": 0.00015, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "15e-5_10240_4_1341", "n_embd": 256}
+{"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_7/attention_kindselective_n_heads4_seed1341", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 4, "protect_bos_token": true, "prevent_from_masking_myself": true, "max_steps": 50000, "warmup_steps": 200, "group": "wider_is_better_7", "use_wandb": true, "kill_self_after_run": false, "random_seed": 1341, "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": 256, "batch_size": 40, "total_batch_size": 10240, "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": true, "disable_selection": false, "mup_enable_coord_check_logging": false, "max_lr": 7e-05, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "7e-5_10240_4_1341", "n_embd": 256}

attention_kindselective_n_heads4_seed1341/dataloader_47500.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55507725e6988f190e4963078652fafa6b68e8d4f79221387612612babf3e1c1
+size 964

attention_kindselective_n_heads4_seed1341/dataloader_49999.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:47776cddb8021172f048a950b83f25b692cb340214b800ce3837c15ceb58907c
+size 964

attention_kindselective_n_heads4_seed1341/log2.txt

@@ -1,605 +1,279 @@
-max_steps: 10000
+max_steps: 50000
+45100 val loss 5.7904
+45100 val perplexity 327.1403
+45100 train 5.674967 (lr=6.0664e-06) (hash(x)=56386155)
 0 val loss 11.3128
 0 val perplexity 81862.0547
-0 val loss 11.3128
-0 val perplexity 81862.0547
-0 train 11.310279 (lr=5.0000e-07) (hash(x)=47078120)
-0 train 11.310279 (lr=7.5000e-07) (hash(x)=47078120)
-100 val loss 9.5732
-100 val perplexity 14373.7412
-100 train 9.498919 (lr=5.0500e-05) (hash(x)=43429388)
-100 val loss 9.1903
-100 val perplexity 9801.1982
-100 train 9.101945 (lr=7.5750e-05) (hash(x)=43429388)
-200 val loss 7.8854
-200 val perplexity 2658.3130
-200 train 8.084438 (lr=1.0000e-04) (hash(x)=52929681)
-200 val loss 7.7259
-200 val perplexity 2266.3394
-200 train 7.960315 (lr=1.5000e-04) (hash(x)=52929681)
-300 val loss 7.6535
-300 val perplexity 2107.9868
-300 train 7.672707 (lr=9.9977e-05) (hash(x)=49930367)
-300 val loss 7.6047
-300 val perplexity 2007.5494
-300 train 7.619209 (lr=1.4997e-04) (hash(x)=49930367)
-400 val loss 7.5925
-400 val perplexity 1983.3094
-400 train 7.398674 (lr=9.9908e-05) (hash(x)=48542946)
-400 val loss 7.5667
-400 val perplexity 1932.8328
-400 train 7.361764 (lr=1.4986e-04) (hash(x)=48542946)
-500 val loss 7.5675
-500 val perplexity 1934.3799
-500 train 7.779499 (lr=9.9792e-05) (hash(x)=55286048)
-500 val loss 7.5630
-500 val perplexity 1925.6095
-500 train 7.758842 (lr=1.4969e-04) (hash(x)=55286048)
-600 val loss 7.5288
-600 val perplexity 1860.8977
-600 train 7.785011 (lr=9.9631e-05) (hash(x)=51577760)
-600 val loss 7.4958
-600 val perplexity 1800.3788
-600 train 7.745537 (lr=1.4945e-04) (hash(x)=51577760)
-700 val loss 7.5018
-700 val perplexity 1811.3163
-700 train 7.658567 (lr=9.9423e-05) (hash(x)=57433471)
-700 val loss 7.4999
-700 val perplexity 1807.9424
-700 train 7.649212 (lr=1.4913e-04) (hash(x)=57433471)
-800 val loss 7.4832
-800 val perplexity 1777.9763
-800 train 7.410454 (lr=9.9170e-05) (hash(x)=49799291)
-800 val loss 7.4499
-800 val perplexity 1719.6324
-800 train 7.368926 (lr=1.4876e-04) (hash(x)=49799291)
-900 val loss 7.4501
-900 val perplexity 1720.0482
-900 train 7.457055 (lr=9.8872e-05) (hash(x)=49502839)
-900 val loss 7.5314
-900 val perplexity 1865.6520
-900 train 7.506033 (lr=1.4831e-04) (hash(x)=49502839)
-1000 val loss 7.4146
-1000 val perplexity 1660.0187
-1000 train 7.690384 (lr=9.8528e-05) (hash(x)=51142904)
-1000 val loss 7.3782
-1000 val perplexity 1600.7816
-1000 train 7.668102 (lr=1.4779e-04) (hash(x)=51142904)
-1100 val loss 7.4470
-1100 val perplexity 1714.6304
-1100 train 7.517180 (lr=9.8140e-05) (hash(x)=52751086)
-1100 val loss 7.4219
-1100 val perplexity 1672.2264
-1100 train 7.479159 (lr=1.4721e-04) (hash(x)=52751086)
-1200 val loss 7.4048
-1200 val perplexity 1643.8654
-1200 train 7.383606 (lr=9.7708e-05) (hash(x)=51538621)
-1200 val loss 7.4675
-1200 val perplexity 1750.2844
-1200 train 7.464672 (lr=1.4656e-04) (hash(x)=51538621)
-1300 val loss 7.4328
-1300 val perplexity 1690.4801
-1300 train 7.501766 (lr=9.7231e-05) (hash(x)=52034040)
-1300 val loss 7.4335
-1300 val perplexity 1691.7429
-1300 train 7.506126 (lr=1.4585e-04) (hash(x)=52034040)
-1400 val loss 7.3738
-1400 val perplexity 1593.6719
-1400 train 7.405330 (lr=9.6711e-05) (hash(x)=50640105)
-1400 val loss 7.3937
-1400 val perplexity 1625.7281
-1400 train 7.437376 (lr=1.4507e-04) (hash(x)=50640105)
-1500 val loss 7.3438
-1500 val perplexity 1546.6002
-1500 train 7.274390 (lr=9.6149e-05) (hash(x)=49016270)
-1500 val loss 7.3919
-1500 val perplexity 1622.8400
-1500 train 7.324384 (lr=1.4422e-04) (hash(x)=49016270)
-1600 val loss 7.3287
-1600 val perplexity 1523.3356
-1600 train 7.098537 (lr=9.5544e-05) (hash(x)=46100488)
-1600 val loss 7.3507
-1600 val perplexity 1557.3582
-1600 train 7.116087 (lr=1.4332e-04) (hash(x)=46100488)
-1700 val loss 7.2968
-1700 val perplexity 1475.5200
-1700 train 7.335896 (lr=9.4897e-05) (hash(x)=49185350)
-1700 val loss 7.3188
-1700 val perplexity 1508.3812
-1700 train 7.345792 (lr=1.4235e-04) (hash(x)=49185350)
-1800 val loss 7.2690
-1800 val perplexity 1435.0470
-1800 train 7.170056 (lr=9.4209e-05) (hash(x)=48024574)
-1800 val loss 7.3890
-1800 val perplexity 1618.1350
-1800 train 7.309584 (lr=1.4131e-04) (hash(x)=48024574)
-1900 val loss 7.2673
-1900 val perplexity 1432.7244
-1900 train 7.072551 (lr=9.3481e-05) (hash(x)=45823189)
-1900 val loss 7.3569
-1900 val perplexity 1566.9396
-1900 train 7.174069 (lr=1.4022e-04) (hash(x)=45823189)
-2000 val loss 7.2557
-2000 val perplexity 1416.1919
-2000 train 7.095765 (lr=9.2714e-05) (hash(x)=45703932)
-2000 val loss 7.3211
-2000 val perplexity 1511.8766
-2000 train 7.151204 (lr=1.3907e-04) (hash(x)=45703932)
-2100 val loss 7.2316
-2100 val perplexity 1382.3717
-2100 train 7.848174 (lr=9.1908e-05) (hash(x)=58570170)
-2100 val loss 7.2782
-2100 val perplexity 1448.3380
-2100 train 7.880461 (lr=1.3786e-04) (hash(x)=58570170)
-2200 val loss 7.2463
-2200 val perplexity 1402.9296
-2200 train 7.282002 (lr=9.1064e-05) (hash(x)=55262880)
-2200 val loss 7.3127
-2200 val perplexity 1499.2522
-2200 train 7.381617 (lr=1.3660e-04) (hash(x)=55262880)
-2300 val loss 7.2115
-2300 val perplexity 1354.9473
-2300 train 6.994785 (lr=9.0182e-05) (hash(x)=46415497)
-2300 val loss 7.3387
-2300 val perplexity 1538.6610
-2300 train 7.105153 (lr=1.3527e-04) (hash(x)=46415497)
-2400 val loss 7.2110
-2400 val perplexity 1354.1884
-2400 train 7.134870 (lr=8.9265e-05) (hash(x)=49272278)
-2400 val loss 7.2934
-2400 val perplexity 1470.6053
-2400 train 7.216728 (lr=1.3390e-04) (hash(x)=49272278)
-2500 val loss 7.2087
-2500 val perplexity 1351.0995
-2500 train 7.070130 (lr=8.8313e-05) (hash(x)=48390803)
-2500 val loss 7.2697
-2500 val perplexity 1436.1204
-2500 train 7.128060 (lr=1.3247e-04) (hash(x)=48390803)
-2600 val loss 7.1845
-2600 val perplexity 1318.7677
-2600 train 7.084837 (lr=8.7326e-05) (hash(x)=47450116)
-2600 val loss 7.2605
-2600 val perplexity 1422.9460
-2600 train 7.147930 (lr=1.3099e-04) (hash(x)=47450116)
-2700 val loss 7.1817
-2700 val perplexity 1315.0879
-2700 train 7.279315 (lr=8.6306e-05) (hash(x)=52681152)
-2700 val loss 7.2477
-2700 val perplexity 1404.8870
-2700 train 7.325132 (lr=1.2946e-04) (hash(x)=52681152)
-2800 val loss 7.1769
-2800 val perplexity 1308.8188
-2800 train 7.244490 (lr=8.5254e-05) (hash(x)=50664094)
-2800 val loss 7.2615
-2800 val perplexity 1424.3207
-2800 train 7.307585 (lr=1.2788e-04) (hash(x)=50664094)
-2900 val loss 7.1611
-2900 val perplexity 1288.2885
-2900 train 6.958514 (lr=8.4170e-05) (hash(x)=47067144)
-2900 val loss 7.2592
-2900 val perplexity 1421.1722
-2900 train 7.065372 (lr=1.2626e-04) (hash(x)=47067144)
-3000 val loss 7.1606
-3000 val perplexity 1287.6467
-3000 train 6.991482 (lr=8.3057e-05) (hash(x)=45015009)
-3000 val loss 7.2889
-3000 val perplexity 1463.9119
-3000 train 7.144633 (lr=1.2459e-04) (hash(x)=45015009)
-3100 val loss 7.1611
-3100 val perplexity 1288.3315
-3100 train 6.985883 (lr=8.1915e-05) (hash(x)=45245896)
-3100 val loss 7.2405
-3100 val perplexity 1394.7473
-3100 train 7.076959 (lr=1.2287e-04) (hash(x)=45245896)
-3200 val loss 7.1369
-3200 val perplexity 1257.4741
-3200 train 7.109494 (lr=8.0745e-05) (hash(x)=49995942)
-3200 val loss 7.2473
-3200 val perplexity 1404.3672
-3200 train 7.197072 (lr=1.2112e-04) (hash(x)=49995942)
-3300 val loss 7.1292
-3300 val perplexity 1247.8973
-3300 train 7.030590 (lr=7.9549e-05) (hash(x)=52311504)
-3300 val loss 7.2531
-3300 val perplexity 1412.5204
-3300 train 7.145674 (lr=1.1932e-04) (hash(x)=52311504)
-3400 val loss 7.1098
-3400 val perplexity 1223.9626
-3400 train 7.093163 (lr=7.8328e-05) (hash(x)=44332917)
-3400 val loss 7.2341
-3400 val perplexity 1385.8531
-3400 train 7.218932 (lr=1.1749e-04) (hash(x)=44332917)
-3500 val loss 7.0946
-3500 val perplexity 1205.4329
-3500 train 7.191734 (lr=7.7082e-05) (hash(x)=56517159)
-3500 val loss 7.2089
-3500 val perplexity 1351.3900
-3500 train 7.302364 (lr=1.1562e-04) (hash(x)=56517159)
-3600 val loss 7.0934
-3600 val perplexity 1203.9680
-3600 train 6.976827 (lr=7.5814e-05) (hash(x)=50720920)
-3600 val loss 7.1955
-3600 val perplexity 1333.4720
-3600 train 7.126245 (lr=1.1372e-04) (hash(x)=50720920)
-3700 val loss 7.0766
-3700 val perplexity 1183.9177
-3700 train 7.557721 (lr=7.4525e-05) (hash(x)=62727701)
-3700 val loss 7.1810
-3700 val perplexity 1314.2090
-3700 train 7.666829 (lr=1.1179e-04) (hash(x)=62727701)
-3800 val loss 7.0697
-3800 val perplexity 1175.7523
-3800 train 6.921579 (lr=7.3215e-05) (hash(x)=54772539)
-3800 val loss 7.1770
-3800 val perplexity 1308.9811
-3800 train 7.023198 (lr=1.0982e-04) (hash(x)=54772539)
-3900 val loss 7.0414
-3900 val perplexity 1143.0204
-3900 train 7.047662 (lr=7.1887e-05) (hash(x)=52274485)
-3900 val loss 7.1482
-3900 val perplexity 1271.7867
-3900 train 7.187829 (lr=1.0783e-04) (hash(x)=52274485)
-4000 val loss 7.0295
-4000 val perplexity 1129.4739
-4000 train 6.761575 (lr=7.0541e-05) (hash(x)=50118307)
-4000 val loss 7.1323
-4000 val perplexity 1251.7776
-4000 train 6.888856 (lr=1.0581e-04) (hash(x)=50118307)
-4100 val loss 7.0174
-4100 val perplexity 1115.9180
-4100 train 6.614292 (lr=6.9180e-05) (hash(x)=42771647)
-4100 val loss 7.1214
-4100 val perplexity 1238.1371
-4100 train 6.728082 (lr=1.0377e-04) (hash(x)=42771647)
-4200 val loss 7.0186
-4200 val perplexity 1117.2550
-4200 train 7.099565 (lr=6.7804e-05) (hash(x)=51748836)
-4200 val loss 7.1332
-4200 val perplexity 1252.8298
-4200 train 7.204681 (lr=1.0171e-04) (hash(x)=51748836)
-4300 val loss 6.9895
-4300 val perplexity 1085.2202
-4300 train 6.994107 (lr=6.6414e-05) (hash(x)=49021280)
-4300 val loss 7.1108
-4300 val perplexity 1225.0831
-4300 train 7.116237 (lr=9.9622e-05) (hash(x)=49021280)
-4400 val loss 6.9606
-4400 val perplexity 1054.2819
-4400 train 6.995679 (lr=6.5013e-05) (hash(x)=55200309)
-4400 val loss 7.0837
-4400 val perplexity 1192.3638
-4400 train 7.194548 (lr=9.7520e-05) (hash(x)=55200309)
-4500 val loss 6.9482
-4500 val perplexity 1041.2869
-4500 train 6.991863 (lr=6.3602e-05) (hash(x)=52085049)
-4500 val loss 7.0670
-4500 val perplexity 1172.6382
-4500 train 7.126850 (lr=9.5403e-05) (hash(x)=52085049)
-4600 val loss 6.9366
-4600 val perplexity 1029.2198
-4600 train 6.905471 (lr=6.2182e-05) (hash(x)=48935595)
-4600 val loss 7.0682
-4600 val perplexity 1174.0856
-4600 train 7.026580 (lr=9.3273e-05) (hash(x)=48935595)
-4700 val loss 6.9126
-4700 val perplexity 1004.8536
-4700 train 7.100014 (lr=6.0754e-05) (hash(x)=49182380)
-4700 val loss 7.0405
-4700 val perplexity 1142.0006
-4700 train 7.225099 (lr=9.1132e-05) (hash(x)=49182380)
-4800 val loss 6.8863
-4800 val perplexity 978.8098
-4800 train 6.715351 (lr=5.9321e-05) (hash(x)=43941929)
-4800 val loss 7.0387
-4800 val perplexity 1139.8903
-4800 train 6.857785 (lr=8.8982e-05) (hash(x)=43941929)
-4900 val loss 6.8664
-4900 val perplexity 959.5104
-4900 train 7.013008 (lr=5.7883e-05) (hash(x)=51852773)
-4900 val loss 7.0199
-4900 val perplexity 1118.6863
-4900 train 7.168573 (lr=8.6825e-05) (hash(x)=51852773)
-5000 val loss 6.8542
-5000 val perplexity 947.8195
-5000 train 6.513768 (lr=5.6442e-05) (hash(x)=40509616)
-5000 val loss 7.0226
-5000 val perplexity 1121.7026
-5000 train 6.697385 (lr=8.4663e-05) (hash(x)=40509616)
-5100 val loss 6.8343
-5100 val perplexity 929.2079
-5100 train 7.078258 (lr=5.5000e-05) (hash(x)=57585369)
-5200 val loss 6.8343
-5200 val perplexity 929.1826
-5200 train 6.838453 (lr=5.3558e-05) (hash(x)=51042313)
-5100 val loss 6.9827
-5100 val perplexity 1077.8414
-5100 train 7.263164 (lr=8.2500e-05) (hash(x)=57585369)
-5300 val loss 6.8171
-5300 val perplexity 913.2907
-5300 train 6.965517 (lr=5.2117e-05) (hash(x)=52001684)
-5200 val loss 6.9860
-5200 val perplexity 1081.3997
-5200 train 7.000556 (lr=8.0337e-05) (hash(x)=51042313)
-5400 val loss 6.7965
-5400 val perplexity 894.7161
-5400 train 6.716750 (lr=5.0679e-05) (hash(x)=48831647)
-5300 val loss 6.9770
-5300 val perplexity 1071.6583
-5300 train 7.099309 (lr=7.8175e-05) (hash(x)=52001684)
-5500 val loss 6.7844
-5500 val perplexity 883.9858
-5500 train 7.172401 (lr=4.9246e-05) (hash(x)=50192069)
-5400 val loss 6.9591
-5400 val perplexity 1052.6428
-5400 train 6.865604 (lr=7.6018e-05) (hash(x)=48831647)
-5600 val loss 6.7744
-5600 val perplexity 875.1322
-5600 train 6.604568 (lr=4.7818e-05) (hash(x)=47208852)
-5700 val loss 6.7622
-5700 val perplexity 864.5762
-5700 train 6.363192 (lr=4.6398e-05) (hash(x)=44061694)
-5500 val loss 6.9728
-5500 val perplexity 1067.2502
-5500 train 7.322748 (lr=7.3868e-05) (hash(x)=50192069)
-5800 val loss 6.7466
-5800 val perplexity 851.1279
-5800 train 7.079750 (lr=4.4987e-05) (hash(x)=56513279)
-5600 val loss 6.9668
-5600 val perplexity 1060.7931
-5600 train 6.799178 (lr=7.1727e-05) (hash(x)=47208852)
-5900 val loss 6.7437
-5900 val perplexity 848.7097
-5900 train 6.870073 (lr=4.3586e-05) (hash(x)=50412818)
-6000 val loss 6.7253
-6000 val perplexity 833.2543
-6000 train 6.548875 (lr=4.2196e-05) (hash(x)=47159634)
-5700 val loss 6.9444
-5700 val perplexity 1037.3643
-5700 train 6.551430 (lr=6.9597e-05) (hash(x)=44061694)
-6100 val loss 6.7193
-6100 val perplexity 828.2003
-6100 train 6.803603 (lr=4.0820e-05) (hash(x)=54312795)
-5800 val loss 6.9374
-5800 val perplexity 1030.0815
-5800 train 7.242342 (lr=6.7480e-05) (hash(x)=56513279)
-6200 val loss 6.7012
-6200 val perplexity 813.3604
-6200 train 6.809578 (lr=3.9459e-05) (hash(x)=54187587)
-6300 val loss 6.6842
-6300 val perplexity 799.6698
-6300 train 6.783376 (lr=3.8113e-05) (hash(x)=53620387)
-5900 val loss 6.9392
-5900 val perplexity 1031.9655
-5900 train 7.063788 (lr=6.5378e-05) (hash(x)=50412818)
-6400 val loss 6.6775
-6400 val perplexity 794.3405
-6400 train 6.623678 (lr=3.6785e-05) (hash(x)=48761774)
-6000 val loss 6.9245
-6000 val perplexity 1016.9347
-6000 train 6.757600 (lr=6.3294e-05) (hash(x)=47159634)
-6500 val loss 6.6764
-6500 val perplexity 793.4588
-6500 train 6.905094 (lr=3.5475e-05) (hash(x)=56690281)
-6600 val loss 6.6623
-6600 val perplexity 782.3674
-6600 train 6.477262 (lr=3.4186e-05) (hash(x)=42985269)
-6100 val loss 6.9089
-6100 val perplexity 1001.1165
-6100 train 6.999359 (lr=6.1230e-05) (hash(x)=54312795)
-6700 val loss 6.6706
-6700 val perplexity 788.8965
-6700 train 6.703038 (lr=3.2918e-05) (hash(x)=53315447)
-6200 val loss 6.8981
-6200 val perplexity 990.4298
-6200 train 6.998672 (lr=5.9188e-05) (hash(x)=54187587)
-6800 val loss 6.6491
-6800 val perplexity 772.0728
-6800 train 7.085675 (lr=3.1672e-05) (hash(x)=61577166)
-6300 val loss 6.8949
-6300 val perplexity 987.1915
-6300 train 6.978578 (lr=5.7169e-05) (hash(x)=53620387)
-6900 val loss 6.6391
-6900 val perplexity 764.3707
-6900 train 6.769608 (lr=3.0451e-05) (hash(x)=54641005)
-6400 val loss 6.8841
-6400 val perplexity 976.5771
-6400 train 6.847974 (lr=5.5177e-05) (hash(x)=48761774)
-7000 val loss 6.6441
-7000 val perplexity 768.2740
-7000 train 7.154501 (lr=2.9255e-05) (hash(x)=60579512)
-6500 val loss 6.8706
-6500 val perplexity 963.5607
-6500 train 7.109013 (lr=5.3213e-05) (hash(x)=56690281)
-7100 val loss 6.6369
-7100 val perplexity 762.7010
-7100 train 6.517390 (lr=2.8085e-05) (hash(x)=53151549)
-6600 val loss 6.8842
-6600 val perplexity 976.7187
-6600 train 6.698448 (lr=5.1279e-05) (hash(x)=42985269)
-7200 val loss 6.6308
-7200 val perplexity 758.0944
-7200 train 7.525359 (lr=2.6943e-05) (hash(x)=71842455)
-6700 val loss 6.8718
-6700 val perplexity 964.7110
-6700 train 6.879877 (lr=4.9377e-05) (hash(x)=53315447)
-7300 val loss 6.6313
-7300 val perplexity 758.4646
-7300 train 6.394084 (lr=2.5830e-05) (hash(x)=44516452)
-6800 val loss 6.8668
-6800 val perplexity 959.8815
-6800 train 7.288239 (lr=4.7509e-05) (hash(x)=61577166)
-7400 val loss 6.6277
-7400 val perplexity 755.7649
-7400 train 6.283513 (lr=2.4746e-05) (hash(x)=42667710)
-6900 val loss 6.8566
-6900 val perplexity 950.1427
-6900 train 6.974115 (lr=4.5676e-05) (hash(x)=54641005)
-7500 val loss 6.6187
-7500 val perplexity 748.9662
-7500 train 6.375410 (lr=2.3694e-05) (hash(x)=47050797)
-7000 val loss 6.8533
-7000 val perplexity 946.9720
-7000 train 7.403100 (lr=4.3882e-05) (hash(x)=60579512)
-7600 val loss 6.6143
-7600 val perplexity 745.6476
-7600 train 6.529490 (lr=2.2674e-05) (hash(x)=49785056)
-7100 val loss 6.8499
-7100 val perplexity 943.8182
-7100 train 6.752803 (lr=4.2128e-05) (hash(x)=53151549)
-7700 val loss 6.6066
-7700 val perplexity 739.9324
-7700 train 6.423981 (lr=2.1687e-05) (hash(x)=53232030)
-7200 val loss 6.8454
-7200 val perplexity 939.5377
-7200 train 7.757356 (lr=4.0414e-05) (hash(x)=71842455)
-7800 val loss 6.6020
-7800 val perplexity 736.5667
-7800 train 6.429112 (lr=2.0735e-05) (hash(x)=48049749)
-7300 val loss 6.8460
-7300 val perplexity 940.1261
-7300 train 6.590199 (lr=3.8745e-05) (hash(x)=44516452)
-7900 val loss 6.5959
-7900 val perplexity 732.0680
-7900 train 6.424716 (lr=1.9818e-05) (hash(x)=44768513)
-7400 val loss 6.8429
-7400 val perplexity 937.2078
-7400 train 6.545752 (lr=3.7120e-05) (hash(x)=42667710)
-8000 val loss 6.5879
-8000 val perplexity 726.2432
-8000 train 6.445358 (lr=1.8936e-05) (hash(x)=46228039)
-7500 val loss 6.8415
-7500 val perplexity 935.8944
-7500 train 6.606590 (lr=3.5541e-05) (hash(x)=47050797)
-8100 val loss 6.5849
-8100 val perplexity 724.0879
-8100 train 6.938977 (lr=1.8092e-05) (hash(x)=60017091)
-7600 val loss 6.8360
-7600 val perplexity 930.7227
-7600 train 6.745418 (lr=3.4011e-05) (hash(x)=49785056)
-8200 val loss 6.5725
-8200 val perplexity 715.1407
-8200 train 6.572504 (lr=1.7286e-05) (hash(x)=49910198)
-7700 val loss 6.8339
-7700 val perplexity 928.8189
-7700 train 6.665596 (lr=3.2531e-05) (hash(x)=53232030)
-8300 val loss 6.5688
-8300 val perplexity 712.4966
-8300 train 6.899169 (lr=1.6519e-05) (hash(x)=57919055)
-7800 val loss 6.8259
-7800 val perplexity 921.4221
-7800 train 6.658208 (lr=3.1102e-05) (hash(x)=48049749)
-8400 val loss 6.5693
-8400 val perplexity 712.8755
-8400 train 6.650244 (lr=1.5791e-05) (hash(x)=49694964)
-7900 val loss 6.8192
-7900 val perplexity 915.2739
-7900 train 6.673261 (lr=2.9726e-05) (hash(x)=44768513)
-8500 val loss 6.5605
-8500 val perplexity 706.6017
-8500 train 6.573118 (lr=1.5103e-05) (hash(x)=53762585)
-8000 val loss 6.8057
-8000 val perplexity 902.9454
-8000 train 6.683189 (lr=2.8405e-05) (hash(x)=46228039)
-8600 val loss 6.5627
-8600 val perplexity 708.1493
-8600 train 6.588408 (lr=1.4456e-05) (hash(x)=51166973)
-8100 val loss 6.8030
-8100 val perplexity 900.5414
-8100 train 7.166605 (lr=2.7138e-05) (hash(x)=60017091)
-8700 val loss 6.5562
-8700 val perplexity 703.6128
-8700 train 6.634116 (lr=1.3851e-05) (hash(x)=53968049)
-8200 val loss 6.7974
-8200 val perplexity 895.5070
-8200 train 6.768152 (lr=2.5929e-05) (hash(x)=49910198)
-8800 val loss 6.5550
-8800 val perplexity 702.7722
-8800 train 6.630794 (lr=1.3289e-05) (hash(x)=59231056)
-8300 val loss 6.7952
-8300 val perplexity 893.5653
-8300 train 7.113091 (lr=2.4778e-05) (hash(x)=57919055)
-8900 val loss 6.5495
-8900 val perplexity 698.9233
-8900 train 6.447311 (lr=1.2769e-05) (hash(x)=50488048)
-8400 val loss 6.7929
-8400 val perplexity 891.5063
-8400 train 6.862132 (lr=2.3686e-05) (hash(x)=49694964)
-9000 val loss 6.5469
-9000 val perplexity 697.0880
-9000 train 6.262195 (lr=1.2292e-05) (hash(x)=44492956)
-8500 val loss 6.7898
-8500 val perplexity 888.7605
-8500 train 6.801354 (lr=2.2655e-05) (hash(x)=53762585)
-9100 val loss 6.5487
-9100 val perplexity 698.3280
-9100 train 6.561413 (lr=1.1860e-05) (hash(x)=51134989)
-8600 val loss 6.7904
-8600 val perplexity 889.2654
-8600 train 6.836792 (lr=2.1685e-05) (hash(x)=51166973)
-9200 val loss 6.5463
-9200 val perplexity 696.6461
-9200 train 6.356756 (lr=1.1472e-05) (hash(x)=48636056)
-8700 val loss 6.7845
-8700 val perplexity 884.0579
-8700 train 6.849996 (lr=2.0777e-05) (hash(x)=53968049)
-9300 val loss 6.5462
-9300 val perplexity 696.5660
-9300 train 6.508790 (lr=1.1128e-05) (hash(x)=50200551)
-8800 val loss 6.7805
-8800 val perplexity 880.4664
-8800 train 6.870674 (lr=1.9933e-05) (hash(x)=59231056)
-9400 val loss 6.5434
-9400 val perplexity 694.6552
-9400 train 6.368688 (lr=1.0830e-05) (hash(x)=48057228)
-8900 val loss 6.7800
-8900 val perplexity 880.0626
-8900 train 6.679636 (lr=1.9153e-05) (hash(x)=50488048)
-9500 val loss 6.5405
-9500 val perplexity 692.6389
-9500 train 6.315319 (lr=1.0577e-05) (hash(x)=48125171)
-9000 val loss 6.7741
-9000 val perplexity 874.9019
-9000 train 6.519008 (lr=1.8439e-05) (hash(x)=44492956)
-9600 val loss 6.5371
-9600 val perplexity 690.2633
-9600 train 6.464390 (lr=1.0369e-05) (hash(x)=53375853)
-9100 val loss 6.7805
-9100 val perplexity 880.5063
-9100 train 6.802141 (lr=1.7790e-05) (hash(x)=51134989)
-9700 val loss 6.5335
-9700 val perplexity 687.7798
-9200 val loss 6.7770
-9200 val perplexity 877.4563
-9700 train 7.310437 (lr=1.0208e-05) (hash(x)=53924631)
-9200 train 6.609503 (lr=1.7208e-05) (hash(x)=48636056)
-9300 val loss 6.7817
-9300 val perplexity 881.5696
-9300 train 6.749519 (lr=1.6692e-05) (hash(x)=50200551)
-9800 val loss 6.5263
-9800 val perplexity 682.8500
-9800 train 6.568887 (lr=1.0092e-05) (hash(x)=48895047)
-9400 val loss 6.7782
-9400 val perplexity 878.4648
-9400 train 6.625322 (lr=1.6245e-05) (hash(x)=48057228)
-9900 val loss 6.5213
-9900 val perplexity 679.4442
-9900 train 6.367904 (lr=1.0023e-05) (hash(x)=44269923)
-9500 val loss 6.7747
-9500 val perplexity 875.4478
-9500 train 6.545719 (lr=1.5865e-05) (hash(x)=48125171)
-9999 val loss 6.5264
-9999 val perplexity 682.9314
-9600 val loss 6.7744
-9600 val perplexity 875.1113
-9600 train 6.720650 (lr=1.5554e-05) (hash(x)=53375853)
-9700 val loss 6.7646
-9700 val perplexity 866.5854
-9700 train 7.470971 (lr=1.5312e-05) (hash(x)=53924631)
-9800 val loss 6.7609
-9800 val perplexity 863.4132
-9800 train 6.783375 (lr=1.5139e-05) (hash(x)=48895047)
-9900 val loss 6.7586
-9900 val perplexity 861.4396
-9900 train 6.626097 (lr=1.5035e-05) (hash(x)=44269923)
-9999 val loss 6.7666
-9999 val perplexity 868.3169
+45200 val loss 5.7941
+45200 val perplexity 328.3590
+45200 train 5.824661 (lr=6.0237e-06) (hash(x)=53389864)
+45300 val loss 5.7925
+45300 val perplexity 327.8424
+45300 train 5.834869 (lr=5.9818e-06) (hash(x)=50817456)
+0 train 11.310279 (lr=3.5000e-07) (hash(x)=47078120)
+45400 val loss 5.7922
+45400 val perplexity 327.7464
+45400 train 5.832580 (lr=5.9407e-06) (hash(x)=50081660)
+100 val loss 9.7348
+100 val perplexity 16895.5098
+100 train 9.668372 (lr=3.5350e-05) (hash(x)=43429388)
+45500 val loss 5.7909
+45500 val perplexity 327.2947
+45500 train 5.550583 (lr=5.9005e-06) (hash(x)=55567121)
+45600 val loss 5.7850
+45600 val perplexity 325.3777
+45600 train 5.868555 (lr=5.8612e-06) (hash(x)=48902854)
+200 val loss 8.2779
+200 val perplexity 3935.9905
+200 train 8.438046 (lr=7.0000e-05) (hash(x)=52929681)
+45700 val loss 5.7834
+45700 val perplexity 324.8476
+45700 train 6.019247 (lr=5.8227e-06) (hash(x)=53803301)
+300 val loss 7.7389
+300 val perplexity 2295.8757
+300 train 7.749171 (lr=6.9999e-05) (hash(x)=49930367)
+45800 val loss 5.7814
+45800 val perplexity 324.2190
+45800 train 5.843807 (lr=5.7851e-06) (hash(x)=47576606)
+400 val loss 7.6187
+400 val perplexity 2035.9438
+400 train 7.430635 (lr=6.9997e-05) (hash(x)=48542946)
+45900 val loss 5.7793
+45900 val perplexity 323.5269
+45900 train 5.661962 (lr=5.7484e-06) (hash(x)=56276981)
+500 val loss 7.5611
+500 val perplexity 1921.9686
+500 train 7.778275 (lr=6.9994e-05) (hash(x)=55286048)
+46000 val loss 5.7781
+46000 val perplexity 323.1599
+46000 train 5.741266 (lr=5.7125e-06) (hash(x)=52174902)
+600 val loss 7.5338
+600 val perplexity 1870.2159
+600 train 7.778021 (lr=6.9990e-05) (hash(x)=51577760)
+46100 val loss 5.7794
+46100 val perplexity 323.5764
+46100 train 5.910207 (lr=5.6775e-06) (hash(x)=52155490)
+700 val loss 7.4794
+700 val perplexity 1771.2211
+700 train 7.640063 (lr=6.9984e-05) (hash(x)=57433471)
+46200 val loss 5.7787
+46200 val perplexity 323.3492
+46200 train 5.635107 (lr=5.6434e-06) (hash(x)=46489984)
+800 val loss 7.4539
+800 val perplexity 1726.5818
+800 train 7.369297 (lr=6.9977e-05) (hash(x)=49799291)
+46300 val loss 5.7793
+46300 val perplexity 323.5242
+46300 train 5.750629 (lr=5.6101e-06) (hash(x)=56603976)
+900 val loss 7.4260
+900 val perplexity 1679.1525
+900 train 7.442145 (lr=6.9969e-05) (hash(x)=49502839)
+46400 val loss 5.7772
+46400 val perplexity 322.8646
+46400 train 5.591374 (lr=5.5777e-06) (hash(x)=51228412)
+46500 val loss 5.7769
+46500 val perplexity 322.7455
+46500 train 5.628456 (lr=5.5462e-06) (hash(x)=48511186)
+1000 val loss 7.3913
+1000 val perplexity 1621.8335
+1000 train 7.669879 (lr=6.9960e-05) (hash(x)=51142904)
+46600 val loss 5.7759
+46600 val perplexity 322.4206
+46600 train 5.813951 (lr=5.5156e-06) (hash(x)=48832039)
+1100 val loss 7.3607
+1100 val perplexity 1572.9337
+1100 train 7.425812 (lr=6.9949e-05) (hash(x)=52751086)
+46700 val loss 5.7767
+46700 val perplexity 322.6856
+46700 train 6.470688 (lr=5.4858e-06) (hash(x)=65821271)
+1200 val loss 7.3503
+1200 val perplexity 1556.7256
+1200 train 7.339065 (lr=6.9937e-05) (hash(x)=51538621)
+46800 val loss 5.7767
+46800 val perplexity 322.6765
+46800 train 5.485295 (lr=5.4569e-06) (hash(x)=48005593)
+1300 val loss 7.3312
+1300 val perplexity 1527.2528
+1300 train 7.411671 (lr=6.9924e-05) (hash(x)=52034040)
+46900 val loss 5.7753
+46900 val perplexity 322.2317
+46900 train 6.050996 (lr=5.4289e-06) (hash(x)=53182453)
+1400 val loss 7.3248
+1400 val perplexity 1517.4226
+1400 train 7.362967 (lr=6.9910e-05) (hash(x)=50640105)
+47000 val loss 5.7761
+47000 val perplexity 322.4943
+47000 train 5.677234 (lr=5.4017e-06) (hash(x)=49385983)
+1500 val loss 7.3135
+1500 val perplexity 1500.4502
+1500 train 7.252579 (lr=6.9894e-05) (hash(x)=49016270)
+47100 val loss 5.7766
+47100 val perplexity 322.6584
+47100 train 5.769415 (lr=5.3755e-06) (hash(x)=48801622)
+1600 val loss 7.2885
+1600 val perplexity 1463.3402
+1600 train 7.062971 (lr=6.9877e-05) (hash(x)=46100488)
+47200 val loss 5.7778
+47200 val perplexity 323.0544
+47200 train 5.511658 (lr=5.3501e-06) (hash(x)=45216690)
+1700 val loss 7.2714
+1700 val perplexity 1438.6268
+1700 train 7.318298 (lr=6.9859e-05) (hash(x)=49185350)
+47300 val loss 5.7776
+47300 val perplexity 322.9699
+47300 train 5.592494 (lr=5.3256e-06) (hash(x)=51185601)
+47400 val loss 5.7751
+47400 val perplexity 322.1754
+47400 train 5.482538 (lr=5.3020e-06) (hash(x)=47743508)
+1800 val loss 7.2397
+1800 val perplexity 1393.7301
+1800 train 7.135669 (lr=6.9840e-05) (hash(x)=48024574)
+47500 val loss 5.7803
+47500 val perplexity 323.8446
+47500 train 5.554578 (lr=5.2792e-06) (hash(x)=49318833)
+1900 val loss 7.2352
+1900 val perplexity 1387.3793
+1900 train 7.070409 (lr=6.9819e-05) (hash(x)=45823189)
+47600 val loss 5.7810
+47600 val perplexity 324.0983
+47600 train 5.741845 (lr=5.2574e-06) (hash(x)=50430826)
+2000 val loss 7.2296
+2000 val perplexity 1379.6533
+2000 train 7.069522 (lr=6.9797e-05) (hash(x)=45703932)
+47700 val loss 5.7792
+47700 val perplexity 323.5000
+47700 train 5.718388 (lr=5.2364e-06) (hash(x)=51966931)
+47800 val loss 5.7772
+47800 val perplexity 322.8683
+47800 train 5.552941 (lr=5.2163e-06) (hash(x)=49517209)
+2100 val loss 7.2192
+2100 val perplexity 1365.4330
+2100 train 7.872229 (lr=6.9774e-05) (hash(x)=58570170)
+47900 val loss 5.7794
+47900 val perplexity 323.5654
+47900 train 5.597253 (lr=5.1972e-06) (hash(x)=49749702)
+48000 val loss 5.7788
+48000 val perplexity 323.3792
+48000 train 5.680276 (lr=5.1788e-06) (hash(x)=52535448)
+2200 val loss 7.2430
+2200 val perplexity 1398.3040
+2200 train 7.315641 (lr=6.9750e-05) (hash(x)=55262880)
+48100 val loss 5.7743
+48100 val perplexity 321.9077
+48100 train 6.455255 (lr=5.1614e-06) (hash(x)=53739181)
+48200 val loss 5.7725
+48200 val perplexity 321.3269
+48200 train 5.751445 (lr=5.1449e-06) (hash(x)=48086710)
+2300 val loss 7.2406
+2300 val perplexity 1394.9767
+2300 train 6.987216 (lr=6.9724e-05) (hash(x)=46415497)
+48300 val loss 5.7727
+48300 val perplexity 321.4185
+48300 train 5.740582 (lr=5.1293e-06) (hash(x)=43991942)
+2400 val loss 7.2117
+2400 val perplexity 1355.2457
+2400 train 7.150833 (lr=6.9697e-05) (hash(x)=49272278)
+48400 val loss 5.7686
+48400 val perplexity 320.0804
+48400 train 5.642438 (lr=5.1145e-06) (hash(x)=50801906)
+48500 val loss 5.7686
+48500 val perplexity 320.1000
+48500 train 5.825756 (lr=5.1007e-06) (hash(x)=48915599)
+2500 val loss 7.2094
+2500 val perplexity 1352.1377
+2500 train 7.084943 (lr=6.9669e-05) (hash(x)=48390803)
+48600 val loss 5.7668
+48600 val perplexity 319.5191
+48600 train 5.929821 (lr=5.0877e-06) (hash(x)=54450172)
+48700 val loss 5.7680
+48700 val perplexity 319.9124
+48700 train 5.660990 (lr=5.0756e-06) (hash(x)=48755509)
+2600 val loss 7.1807
+2600 val perplexity 1313.8004
+2600 train 7.066188 (lr=6.9640e-05) (hash(x)=47450116)
+48800 val loss 5.7696
+48800 val perplexity 320.4213
+48800 train 5.660839 (lr=5.0644e-06) (hash(x)=51380096)
+2700 val loss 7.1767
+2700 val perplexity 1308.5830
+2700 train 7.263812 (lr=6.9609e-05) (hash(x)=52681152)
+48900 val loss 5.7680
+48900 val perplexity 319.9099
+48900 train 5.630540 (lr=5.0542e-06) (hash(x)=43404218)
+2800 val loss 7.1723
+2800 val perplexity 1302.8984
+2800 train 7.220819 (lr=6.9577e-05) (hash(x)=50664094)
+49000 val loss 5.7681
+49000 val perplexity 319.9189
+49000 train 5.758193 (lr=5.0448e-06) (hash(x)=55059739)
+2900 val loss 7.1650
+2900 val perplexity 1293.3875
+2900 train 6.984344 (lr=6.9544e-05) (hash(x)=47067144)
+49100 val loss 5.7706
+49100 val perplexity 320.7290
+49100 train 5.996271 (lr=5.0363e-06) (hash(x)=47889309)
+3000 val loss 7.1467
+3000 val perplexity 1269.8622
+3000 train 6.990779 (lr=6.9510e-05) (hash(x)=45015009)
+49200 val loss 5.7695
+49200 val perplexity 320.3733
+49200 train 5.512230 (lr=5.0286e-06) (hash(x)=46202589)
+3100 val loss 7.1704
+3100 val perplexity 1300.3754
+3100 train 7.004176 (lr=6.9474e-05) (hash(x)=45245896)
+49300 val loss 5.7690
+49300 val perplexity 320.2130
+49300 train 5.721933 (lr=5.0219e-06) (hash(x)=49161813)
+3200 val loss 7.1561
+3200 val perplexity 1281.9510
+3200 train 7.116989 (lr=6.9438e-05) (hash(x)=49995942)
+49400 val loss 5.7681
+49400 val perplexity 319.9419
+49400 train 5.601577 (lr=5.0161e-06) (hash(x)=49505044)
+3300 val loss 7.1272
+3300 val perplexity 1245.4196
+3300 train 7.024605 (lr=6.9400e-05) (hash(x)=52311504)
+49500 val loss 5.7683
+49500 val perplexity 319.9982
+49500 train 5.544023 (lr=5.0112e-06) (hash(x)=46610327)
+3400 val loss 7.1102
+3400 val perplexity 1224.3795
+3400 train 7.109678 (lr=6.9360e-05) (hash(x)=44332917)
+49600 val loss 5.7680
+49600 val perplexity 319.8917
+49600 train 6.107419 (lr=5.0072e-06) (hash(x)=51188240)
+3500 val loss 7.1105
+3500 val perplexity 1224.7303
+3500 train 7.218296 (lr=6.9320e-05) (hash(x)=56517159)
+49700 val loss 5.7667
+49700 val perplexity 319.4920
+49700 train 5.552507 (lr=5.0040e-06) (hash(x)=41492016)
+3600 val loss 7.0860
+3600 val perplexity 1195.0756
+3600 train 6.962664 (lr=6.9278e-05) (hash(x)=50720920)
+49800 val loss 5.7675
+49800 val perplexity 319.7239
+49800 train 5.705134 (lr=5.0018e-06) (hash(x)=53488833)
+3700 val loss 7.0801
+3700 val perplexity 1188.0885
+3700 train 7.562508 (lr=6.9235e-05) (hash(x)=62727701)
+49900 val loss 5.7662
+49900 val perplexity 319.3073
+49900 train 6.118833 (lr=5.0004e-06) (hash(x)=60260935)
+3800 val loss 7.0837
+3800 val perplexity 1192.3132
+3800 train 6.928957 (lr=6.9191e-05) (hash(x)=54772539)
+49999 val loss 5.7655
+49999 val perplexity 319.1097
+3900 val loss 7.0776
+3900 val perplexity 1185.0813
+3900 train 7.083435 (lr=6.9146e-05) (hash(x)=52274485)
+4000 val loss 7.0939
+4000 val perplexity 1204.5883
+4000 train 6.805799 (lr=6.9099e-05) (hash(x)=50118307)
+4100 val loss 7.0658
+4100 val perplexity 1171.2137
+4100 train 6.668285 (lr=6.9051e-05) (hash(x)=42771647)
+4200 val loss 7.0463
+4200 val perplexity 1148.6185
+4200 train 7.126415 (lr=6.9002e-05) (hash(x)=51748836)

attention_kindselective_n_heads4_seed1341/model_02500.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3e1c2187204d565e2ee51bd04e1f6b5bef5437dd443df8d65d370bbdcdf46c04
+oid sha256:c484aa037e9b115403161e2fe304d0c7a89dc6b79d216e3cdfd363f6ca13d25d
 size 92843394

attention_kindselective_n_heads4_seed1341/model_47500.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6aeb2a7a2351f8f16532e3f96f1535852348a874b6e53498f54dbb2dbea87f2e
+size 92843394

attention_kindselective_n_heads4_seed1341/model_49999.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a27f8d94a91205f44b4097397cbe96f34c7125f7acaf12a27ed77f3f225c47c
+size 92843394

attention_kindselective_n_heads4_seed1341/optimizer_02500.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:914f651e601811726eebb0a1db73bb2e4e98a78ce6c3876e94d1b62ab6a6dda9
+oid sha256:c5a91ef2bc89f32f183c058e5eeb515108d458e3939109aed168281f8270a000
 size 179406214

attention_kindselective_n_heads4_seed1341/optimizer_47500.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:142ca7b4714d69fb474f9d95e2cb07eb349cb18b2cb2b88c2fcae9ebdf2d59c3
+size 179406214

attention_kindselective_n_heads4_seed1341/optimizer_49999.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:853ea0e957cdc71f93c4e51e26d64a9edbc8edb91db6673397e9508908e6913c
+size 179406214