Upload folder using huggingface_hub
Browse files
4x_smaller_bs_half_lr_half_seq_len/args.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"hellaswag": true, "attention_kind": "selective", "log_dir": "4x_smaller_bs_half_lr_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": 131072, "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": 1.5e-05, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": null, "n_embd": 768}
|
4x_smaller_bs_half_lr_half_seq_len/dataloader_02499.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dc64603b8ea6a36a9ccabc0a5d7c7479b649579c80e31622d3273d19aa5ac2a8
|
| 3 |
+
size 964
|
4x_smaller_bs_half_lr_half_seq_len/log2.txt
ADDED
|
@@ -0,0 +1,303 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
max_steps: 2500
|
| 2 |
+
0 val loss 10.9883
|
| 3 |
+
0 val perplexity 59177.6016
|
| 4 |
+
0 train 10.987196 (lr=2.0979e-08) (hash(x)=44290605)
|
| 5 |
+
10 train 10.983902 (lr=2.3077e-07) (hash(x)=38564062)
|
| 6 |
+
20 train 10.970700 (lr=4.4056e-07) (hash(x)=42646055)
|
| 7 |
+
30 train 10.951057 (lr=6.5035e-07) (hash(x)=38523121)
|
| 8 |
+
40 train 10.915579 (lr=8.6014e-07) (hash(x)=34789261)
|
| 9 |
+
50 train 10.879971 (lr=1.0699e-06) (hash(x)=35228141)
|
| 10 |
+
60 train 10.833032 (lr=1.2797e-06) (hash(x)=39587191)
|
| 11 |
+
70 train 10.774123 (lr=1.4895e-06) (hash(x)=39506396)
|
| 12 |
+
80 train 10.725358 (lr=1.6993e-06) (hash(x)=44465412)
|
| 13 |
+
90 train 10.638872 (lr=1.9091e-06) (hash(x)=46934196)
|
| 14 |
+
100 val loss 10.5456
|
| 15 |
+
100 val perplexity 38009.1719
|
| 16 |
+
100 train 10.548512 (lr=2.1189e-06) (hash(x)=37401443)
|
| 17 |
+
110 train 10.453033 (lr=2.3287e-06) (hash(x)=36880637)
|
| 18 |
+
120 train 10.333361 (lr=2.5385e-06) (hash(x)=40760978)
|
| 19 |
+
130 train 10.240342 (lr=2.7483e-06) (hash(x)=34332830)
|
| 20 |
+
140 train 10.125100 (lr=2.9580e-06) (hash(x)=41010950)
|
| 21 |
+
150 train 10.016959 (lr=3.1678e-06) (hash(x)=41413911)
|
| 22 |
+
160 train 9.938765 (lr=3.3776e-06) (hash(x)=43969526)
|
| 23 |
+
170 train 9.858195 (lr=3.5874e-06) (hash(x)=36475394)
|
| 24 |
+
180 train 9.822458 (lr=3.7972e-06) (hash(x)=34893035)
|
| 25 |
+
190 train 9.778758 (lr=4.0070e-06) (hash(x)=39125577)
|
| 26 |
+
200 val loss 9.7399
|
| 27 |
+
200 val perplexity 16981.8379
|
| 28 |
+
200 train 9.783566 (lr=4.2168e-06) (hash(x)=41789901)
|
| 29 |
+
210 train 9.775907 (lr=4.4266e-06) (hash(x)=38495745)
|
| 30 |
+
220 train 9.710948 (lr=4.6364e-06) (hash(x)=44203053)
|
| 31 |
+
230 train 9.643511 (lr=4.8462e-06) (hash(x)=30353179)
|
| 32 |
+
240 train 9.652902 (lr=5.0559e-06) (hash(x)=36564311)
|
| 33 |
+
250 train 9.703348 (lr=5.2657e-06) (hash(x)=38917268)
|
| 34 |
+
260 train 9.561985 (lr=5.4755e-06) (hash(x)=37263509)
|
| 35 |
+
270 train 9.576735 (lr=5.6853e-06) (hash(x)=36081848)
|
| 36 |
+
280 train 9.536775 (lr=5.8951e-06) (hash(x)=38157778)
|
| 37 |
+
290 train 9.459773 (lr=6.1049e-06) (hash(x)=39695381)
|
| 38 |
+
300 val loss 9.4520
|
| 39 |
+
300 val perplexity 12733.2510
|
| 40 |
+
300 train 9.454054 (lr=6.3147e-06) (hash(x)=38914105)
|
| 41 |
+
310 train 9.422870 (lr=6.5245e-06) (hash(x)=40676791)
|
| 42 |
+
320 train 9.391773 (lr=6.7343e-06) (hash(x)=42043458)
|
| 43 |
+
330 train 9.351944 (lr=6.9441e-06) (hash(x)=43861760)
|
| 44 |
+
340 train 9.303730 (lr=7.1538e-06) (hash(x)=42434651)
|
| 45 |
+
350 train 9.334338 (lr=7.3636e-06) (hash(x)=42878691)
|
| 46 |
+
360 train 9.132668 (lr=7.5734e-06) (hash(x)=37008216)
|
| 47 |
+
370 train 9.063375 (lr=7.7832e-06) (hash(x)=44318637)
|
| 48 |
+
380 train 9.078611 (lr=7.9930e-06) (hash(x)=39642311)
|
| 49 |
+
390 train 8.941845 (lr=8.2028e-06) (hash(x)=38871537)
|
| 50 |
+
400 val loss 8.9100
|
| 51 |
+
400 val perplexity 7405.5044
|
| 52 |
+
400 train 8.915948 (lr=8.4126e-06) (hash(x)=38494942)
|
| 53 |
+
410 train 8.821939 (lr=8.6224e-06) (hash(x)=36423663)
|
| 54 |
+
420 train 8.746850 (lr=8.8322e-06) (hash(x)=37628657)
|
| 55 |
+
430 train 8.667839 (lr=9.0420e-06) (hash(x)=39593287)
|
| 56 |
+
440 train 8.605666 (lr=9.2517e-06) (hash(x)=38939280)
|
| 57 |
+
450 train 8.528643 (lr=9.4615e-06) (hash(x)=41400408)
|
| 58 |
+
460 train 8.522301 (lr=9.6713e-06) (hash(x)=39718278)
|
| 59 |
+
470 train 8.481720 (lr=9.8811e-06) (hash(x)=39685768)
|
| 60 |
+
480 train 8.475543 (lr=1.0091e-05) (hash(x)=37108161)
|
| 61 |
+
490 train 8.344969 (lr=1.0301e-05) (hash(x)=39677447)
|
| 62 |
+
500 val loss 8.2716
|
| 63 |
+
500 val perplexity 3911.3613
|
| 64 |
+
500 train 8.293922 (lr=1.0510e-05) (hash(x)=38805353)
|
| 65 |
+
510 train 8.172845 (lr=1.0720e-05) (hash(x)=37151493)
|
| 66 |
+
520 train 8.160428 (lr=1.0930e-05) (hash(x)=42467708)
|
| 67 |
+
530 train 8.005136 (lr=1.1140e-05) (hash(x)=39283765)
|
| 68 |
+
540 train 8.018361 (lr=1.1350e-05) (hash(x)=36360047)
|
| 69 |
+
550 train 8.012372 (lr=1.1559e-05) (hash(x)=36312866)
|
| 70 |
+
560 train 7.872485 (lr=1.1769e-05) (hash(x)=41515622)
|
| 71 |
+
570 train 7.856777 (lr=1.1979e-05) (hash(x)=36732231)
|
| 72 |
+
580 train 7.792838 (lr=1.2189e-05) (hash(x)=41329361)
|
| 73 |
+
590 train 7.696174 (lr=1.2399e-05) (hash(x)=38604113)
|
| 74 |
+
600 val loss 7.7623
|
| 75 |
+
600 val perplexity 2350.3792
|
| 76 |
+
600 train 7.707370 (lr=1.2608e-05) (hash(x)=48521551)
|
| 77 |
+
610 train 7.790296 (lr=1.2818e-05) (hash(x)=46162817)
|
| 78 |
+
620 train 7.815807 (lr=1.3028e-05) (hash(x)=40949420)
|
| 79 |
+
630 train 7.744004 (lr=1.3238e-05) (hash(x)=41577136)
|
| 80 |
+
640 train 7.595581 (lr=1.3448e-05) (hash(x)=40803922)
|
| 81 |
+
650 train 7.482319 (lr=1.3657e-05) (hash(x)=37788936)
|
| 82 |
+
660 train 7.474987 (lr=1.3867e-05) (hash(x)=39766064)
|
| 83 |
+
670 train 7.486190 (lr=1.4077e-05) (hash(x)=37945924)
|
| 84 |
+
680 train 7.465253 (lr=1.4287e-05) (hash(x)=37529238)
|
| 85 |
+
690 train 7.483014 (lr=1.4497e-05) (hash(x)=44662178)
|
| 86 |
+
700 val loss 7.4751
|
| 87 |
+
700 val perplexity 1763.6294
|
| 88 |
+
700 train 7.442530 (lr=1.4706e-05) (hash(x)=40331394)
|
| 89 |
+
710 train 7.356412 (lr=1.4916e-05) (hash(x)=37685289)
|
| 90 |
+
720 train 7.286057 (lr=1.5000e-05) (hash(x)=36102366)
|
| 91 |
+
730 train 7.324800 (lr=1.4998e-05) (hash(x)=40147452)
|
| 92 |
+
740 train 7.267545 (lr=1.4993e-05) (hash(x)=43905866)
|
| 93 |
+
750 train 7.399191 (lr=1.4987e-05) (hash(x)=42843608)
|
| 94 |
+
760 train 7.409278 (lr=1.4979e-05) (hash(x)=37296224)
|
| 95 |
+
770 train 7.313937 (lr=1.4968e-05) (hash(x)=41230008)
|
| 96 |
+
780 train 7.221914 (lr=1.4956e-05) (hash(x)=36407773)
|
| 97 |
+
790 train 7.376118 (lr=1.4941e-05) (hash(x)=42561722)
|
| 98 |
+
800 val loss 7.3352
|
| 99 |
+
800 val perplexity 1533.2998
|
| 100 |
+
800 train 7.219405 (lr=1.4925e-05) (hash(x)=48340441)
|
| 101 |
+
810 train 7.200429 (lr=1.4906e-05) (hash(x)=36261826)
|
| 102 |
+
820 train 7.126314 (lr=1.4885e-05) (hash(x)=37918881)
|
| 103 |
+
830 train 7.233766 (lr=1.4862e-05) (hash(x)=38608334)
|
| 104 |
+
840 train 7.124430 (lr=1.4837e-05) (hash(x)=39403505)
|
| 105 |
+
850 train 7.176360 (lr=1.4810e-05) (hash(x)=43741441)
|
| 106 |
+
860 train 7.174889 (lr=1.4781e-05) (hash(x)=38030739)
|
| 107 |
+
870 train 7.256120 (lr=1.4750e-05) (hash(x)=36064710)
|
| 108 |
+
880 train 7.313010 (lr=1.4717e-05) (hash(x)=40349968)
|
| 109 |
+
890 train 7.249527 (lr=1.4682e-05) (hash(x)=38715806)
|
| 110 |
+
900 val loss 7.2489
|
| 111 |
+
900 val perplexity 1406.4869
|
| 112 |
+
900 train 7.263391 (lr=1.4645e-05) (hash(x)=35310433)
|
| 113 |
+
910 train 7.183582 (lr=1.4606e-05) (hash(x)=37426246)
|
| 114 |
+
920 train 7.150632 (lr=1.4565e-05) (hash(x)=39966506)
|
| 115 |
+
930 train 7.126366 (lr=1.4522e-05) (hash(x)=37747046)
|
| 116 |
+
940 train 7.163310 (lr=1.4478e-05) (hash(x)=42710584)
|
| 117 |
+
950 train 7.182676 (lr=1.4431e-05) (hash(x)=41617662)
|
| 118 |
+
960 train 7.174756 (lr=1.4382e-05) (hash(x)=40145473)
|
| 119 |
+
970 train 7.098168 (lr=1.4332e-05) (hash(x)=38513059)
|
| 120 |
+
980 train 7.043861 (lr=1.4279e-05) (hash(x)=38574237)
|
| 121 |
+
990 train 7.148112 (lr=1.4225e-05) (hash(x)=37633181)
|
| 122 |
+
1000 val loss 7.1861
|
| 123 |
+
1000 val perplexity 1320.9963
|
| 124 |
+
1000 train 7.011847 (lr=1.4169e-05) (hash(x)=42781027)
|
| 125 |
+
1010 train 7.141905 (lr=1.4110e-05) (hash(x)=48329642)
|
| 126 |
+
1020 train 6.997019 (lr=1.4051e-05) (hash(x)=36577969)
|
| 127 |
+
1030 train 7.045058 (lr=1.3989e-05) (hash(x)=39789757)
|
| 128 |
+
1040 train 6.968916 (lr=1.3926e-05) (hash(x)=43293764)
|
| 129 |
+
1050 train 7.247198 (lr=1.3860e-05) (hash(x)=40077223)
|
| 130 |
+
1060 train 7.111925 (lr=1.3793e-05) (hash(x)=41035822)
|
| 131 |
+
1070 train 7.198096 (lr=1.3725e-05) (hash(x)=42069615)
|
| 132 |
+
1080 train 7.062248 (lr=1.3654e-05) (hash(x)=42668159)
|
| 133 |
+
1090 train 7.042990 (lr=1.3582e-05) (hash(x)=36477372)
|
| 134 |
+
1100 val loss 7.1245
|
| 135 |
+
1100 val perplexity 1242.0374
|
| 136 |
+
1100 train 7.101448 (lr=1.3509e-05) (hash(x)=41357005)
|
| 137 |
+
1110 train 7.115828 (lr=1.3434e-05) (hash(x)=42852875)
|
| 138 |
+
1120 train 7.008416 (lr=1.3357e-05) (hash(x)=35310013)
|
| 139 |
+
1130 train 7.075465 (lr=1.3278e-05) (hash(x)=51231839)
|
| 140 |
+
1140 train 6.985238 (lr=1.3198e-05) (hash(x)=41431568)
|
| 141 |
+
1150 train 7.004578 (lr=1.3117e-05) (hash(x)=41848748)
|
| 142 |
+
1160 train 6.896893 (lr=1.3033e-05) (hash(x)=36702617)
|
| 143 |
+
1170 train 7.079871 (lr=1.2949e-05) (hash(x)=40801937)
|
| 144 |
+
1180 train 7.095430 (lr=1.2863e-05) (hash(x)=41936951)
|
| 145 |
+
1190 train 6.837819 (lr=1.2775e-05) (hash(x)=41913508)
|
| 146 |
+
1200 val loss 7.0866
|
| 147 |
+
1200 val perplexity 1195.8895
|
| 148 |
+
1200 train 6.969180 (lr=1.2687e-05) (hash(x)=36842847)
|
| 149 |
+
1210 train 6.975598 (lr=1.2596e-05) (hash(x)=30499991)
|
| 150 |
+
1220 train 6.957082 (lr=1.2505e-05) (hash(x)=41929560)
|
| 151 |
+
1230 train 7.102791 (lr=1.2412e-05) (hash(x)=47738317)
|
| 152 |
+
1240 train 7.120340 (lr=1.2318e-05) (hash(x)=38256802)
|
| 153 |
+
1250 train 7.078293 (lr=1.2222e-05) (hash(x)=28755106)
|
| 154 |
+
1260 train 7.030218 (lr=1.2126e-05) (hash(x)=38336891)
|
| 155 |
+
1270 train 7.085950 (lr=1.2028e-05) (hash(x)=36084046)
|
| 156 |
+
1280 train 7.075439 (lr=1.1929e-05) (hash(x)=37921865)
|
| 157 |
+
1290 train 6.831306 (lr=1.1829e-05) (hash(x)=38656933)
|
| 158 |
+
1300 val loss 7.0392
|
| 159 |
+
1300 val perplexity 1140.5215
|
| 160 |
+
1300 train 6.900213 (lr=1.1727e-05) (hash(x)=39331872)
|
| 161 |
+
1310 train 6.927700 (lr=1.1625e-05) (hash(x)=35713543)
|
| 162 |
+
1320 train 6.914740 (lr=1.1522e-05) (hash(x)=42221951)
|
| 163 |
+
1330 train 7.036623 (lr=1.1417e-05) (hash(x)=49371177)
|
| 164 |
+
1340 train 6.992999 (lr=1.1312e-05) (hash(x)=39825492)
|
| 165 |
+
1350 train 6.863598 (lr=1.1205e-05) (hash(x)=38122709)
|
| 166 |
+
1360 train 6.947845 (lr=1.1098e-05) (hash(x)=46034654)
|
| 167 |
+
1370 train 6.871446 (lr=1.0990e-05) (hash(x)=36717950)
|
| 168 |
+
1380 train 6.875348 (lr=1.0881e-05) (hash(x)=40410179)
|
| 169 |
+
1390 train 6.896519 (lr=1.0771e-05) (hash(x)=46535981)
|
| 170 |
+
1400 val loss 7.0105
|
| 171 |
+
1400 val perplexity 1108.2163
|
| 172 |
+
1400 train 6.854383 (lr=1.0661e-05) (hash(x)=35556187)
|
| 173 |
+
1410 train 6.907253 (lr=1.0549e-05) (hash(x)=47156509)
|
| 174 |
+
1420 train 7.004330 (lr=1.0437e-05) (hash(x)=44281694)
|
| 175 |
+
1430 train 7.111400 (lr=1.0325e-05) (hash(x)=38833525)
|
| 176 |
+
1440 train 7.146225 (lr=1.0211e-05) (hash(x)=42715085)
|
| 177 |
+
1450 train 7.009366 (lr=1.0097e-05) (hash(x)=42032181)
|
| 178 |
+
1460 train 7.039539 (lr=9.9827e-06) (hash(x)=42012177)
|
| 179 |
+
1470 train 6.960066 (lr=9.8676e-06) (hash(x)=41738021)
|
| 180 |
+
1480 train 6.869416 (lr=9.7520e-06) (hash(x)=48868810)
|
| 181 |
+
1490 train 7.356836 (lr=9.6360e-06) (hash(x)=50198350)
|
| 182 |
+
1500 val loss 6.9743
|
| 183 |
+
1500 val perplexity 1068.7821
|
| 184 |
+
1500 train 6.961478 (lr=9.5195e-06) (hash(x)=44018572)
|
| 185 |
+
1510 train 6.918033 (lr=9.4026e-06) (hash(x)=40114593)
|
| 186 |
+
1520 train 6.839454 (lr=9.2854e-06) (hash(x)=41475808)
|
| 187 |
+
1530 train 6.941401 (lr=9.1678e-06) (hash(x)=42156287)
|
| 188 |
+
1540 train 6.972492 (lr=9.0500e-06) (hash(x)=41903473)
|
| 189 |
+
1550 train 6.897877 (lr=8.9319e-06) (hash(x)=45895413)
|
| 190 |
+
1560 train 6.969629 (lr=8.8136e-06) (hash(x)=56812849)
|
| 191 |
+
1570 train 6.823680 (lr=8.6952e-06) (hash(x)=38509270)
|
| 192 |
+
1580 train 6.828520 (lr=8.5766e-06) (hash(x)=39318554)
|
| 193 |
+
1590 train 6.943930 (lr=8.4579e-06) (hash(x)=43078392)
|
| 194 |
+
1600 val loss 6.9467
|
| 195 |
+
1600 val perplexity 1039.7418
|
| 196 |
+
1600 train 7.069693 (lr=8.3391e-06) (hash(x)=41122655)
|
| 197 |
+
1610 train 6.998374 (lr=8.2203e-06) (hash(x)=36120625)
|
| 198 |
+
1620 train 6.855536 (lr=8.1015e-06) (hash(x)=35765792)
|
| 199 |
+
1630 train 7.095515 (lr=7.9828e-06) (hash(x)=47089042)
|
| 200 |
+
1640 train 6.852379 (lr=7.8641e-06) (hash(x)=50711559)
|
| 201 |
+
1650 train 6.869404 (lr=7.7456e-06) (hash(x)=45815369)
|
| 202 |
+
1660 train 6.730164 (lr=7.6272e-06) (hash(x)=32816402)
|
| 203 |
+
1670 train 6.833857 (lr=7.5090e-06) (hash(x)=37278001)
|
| 204 |
+
1680 train 6.896299 (lr=7.3910e-06) (hash(x)=40915022)
|
| 205 |
+
1690 train 6.734944 (lr=7.2733e-06) (hash(x)=39134730)
|
| 206 |
+
1700 val loss 6.9317
|
| 207 |
+
1700 val perplexity 1024.1831
|
| 208 |
+
1700 train 6.818570 (lr=7.1560e-06) (hash(x)=43322472)
|
| 209 |
+
1710 train 6.880149 (lr=7.0389e-06) (hash(x)=44220064)
|
| 210 |
+
1720 train 6.938684 (lr=6.9222e-06) (hash(x)=38892807)
|
| 211 |
+
1730 train 7.044489 (lr=6.8060e-06) (hash(x)=43115553)
|
| 212 |
+
1740 train 6.879650 (lr=6.6901e-06) (hash(x)=39288672)
|
| 213 |
+
1750 train 6.975536 (lr=6.5748e-06) (hash(x)=41124961)
|
| 214 |
+
1760 train 6.849125 (lr=6.4600e-06) (hash(x)=36048907)
|
| 215 |
+
1770 train 6.849308 (lr=6.3457e-06) (hash(x)=38027887)
|
| 216 |
+
1780 train 6.819469 (lr=6.2320e-06) (hash(x)=38145978)
|
| 217 |
+
1790 train 6.910053 (lr=6.1190e-06) (hash(x)=44351536)
|
| 218 |
+
1800 val loss 6.9073
|
| 219 |
+
1800 val perplexity 999.4962
|
| 220 |
+
1800 train 6.904034 (lr=6.0066e-06) (hash(x)=39518627)
|
| 221 |
+
1810 train 6.755105 (lr=5.8949e-06) (hash(x)=41542967)
|
| 222 |
+
1820 train 6.800942 (lr=5.7839e-06) (hash(x)=38500785)
|
| 223 |
+
1830 train 6.844421 (lr=5.6737e-06) (hash(x)=37888270)
|
| 224 |
+
1840 train 6.719558 (lr=5.5643e-06) (hash(x)=30199535)
|
| 225 |
+
1850 train 6.728424 (lr=5.4558e-06) (hash(x)=40458277)
|
| 226 |
+
1860 train 6.704224 (lr=5.3481e-06) (hash(x)=42984408)
|
| 227 |
+
1870 train 6.698699 (lr=5.2413e-06) (hash(x)=39240321)
|
| 228 |
+
1880 train 6.707630 (lr=5.1354e-06) (hash(x)=37169036)
|
| 229 |
+
1890 train 6.782664 (lr=5.0305e-06) (hash(x)=40156837)
|
| 230 |
+
1900 val loss 6.8976
|
| 231 |
+
1900 val perplexity 989.8618
|
| 232 |
+
1900 train 6.889485 (lr=4.9266e-06) (hash(x)=36032188)
|
| 233 |
+
1910 train 6.965537 (lr=4.8237e-06) (hash(x)=42795519)
|
| 234 |
+
1920 train 6.892735 (lr=4.7219e-06) (hash(x)=35518925)
|
| 235 |
+
1930 train 6.951318 (lr=4.6211e-06) (hash(x)=40438046)
|
| 236 |
+
1940 train 6.762111 (lr=4.5215e-06) (hash(x)=38291653)
|
| 237 |
+
1950 train 6.887587 (lr=4.4231e-06) (hash(x)=42821483)
|
| 238 |
+
1960 train 6.850200 (lr=4.3258e-06) (hash(x)=37781932)
|
| 239 |
+
1970 train 6.939731 (lr=4.2298e-06) (hash(x)=42543275)
|
| 240 |
+
1980 train 6.837874 (lr=4.1350e-06) (hash(x)=37836462)
|
| 241 |
+
1990 train 6.870891 (lr=4.0414e-06) (hash(x)=39708342)
|
| 242 |
+
2000 val loss 6.8812
|
| 243 |
+
2000 val perplexity 973.8149
|
| 244 |
+
2000 train 6.770031 (lr=3.9492e-06) (hash(x)=38050986)
|
| 245 |
+
2010 train 6.963208 (lr=3.8583e-06) (hash(x)=43963198)
|
| 246 |
+
2020 train 6.689958 (lr=3.7688e-06) (hash(x)=35638171)
|
| 247 |
+
2030 train 6.697921 (lr=3.6806e-06) (hash(x)=33906793)
|
| 248 |
+
2040 train 6.709877 (lr=3.5939e-06) (hash(x)=35880906)
|
| 249 |
+
2050 train 6.786647 (lr=3.5086e-06) (hash(x)=34453353)
|
| 250 |
+
2060 train 6.726500 (lr=3.4248e-06) (hash(x)=42567743)
|
| 251 |
+
2070 train 6.740697 (lr=3.3425e-06) (hash(x)=40686891)
|
| 252 |
+
2080 train 6.589478 (lr=3.2617e-06) (hash(x)=33852944)
|
| 253 |
+
2090 train 6.879174 (lr=3.1824e-06) (hash(x)=37270130)
|
| 254 |
+
2100 val loss 6.8731
|
| 255 |
+
2100 val perplexity 965.9142
|
| 256 |
+
2100 train 6.931888 (lr=3.1047e-06) (hash(x)=41991671)
|
| 257 |
+
2110 train 6.881395 (lr=3.0286e-06) (hash(x)=38825122)
|
| 258 |
+
2120 train 6.913153 (lr=2.9542e-06) (hash(x)=44277711)
|
| 259 |
+
2130 train 6.899714 (lr=2.8813e-06) (hash(x)=53809569)
|
| 260 |
+
2140 train 6.776664 (lr=2.8102e-06) (hash(x)=36559771)
|
| 261 |
+
2150 train 6.823159 (lr=2.7407e-06) (hash(x)=40161967)
|
| 262 |
+
2160 train 6.801567 (lr=2.6729e-06) (hash(x)=33539846)
|
| 263 |
+
2170 train 6.866282 (lr=2.6068e-06) (hash(x)=39682712)
|
| 264 |
+
2180 train 6.914186 (lr=2.5425e-06) (hash(x)=35780230)
|
| 265 |
+
2190 train 6.804833 (lr=2.4800e-06) (hash(x)=43980566)
|
| 266 |
+
2200 val loss 6.8645
|
| 267 |
+
2200 val perplexity 957.6674
|
| 268 |
+
2200 train 6.776078 (lr=2.4192e-06) (hash(x)=34935313)
|
| 269 |
+
2210 train 6.824680 (lr=2.3603e-06) (hash(x)=39724733)
|
| 270 |
+
2220 train 7.287607 (lr=2.3032e-06) (hash(x)=43703966)
|
| 271 |
+
2230 train 6.779405 (lr=2.2479e-06) (hash(x)=38399765)
|
| 272 |
+
2240 train 6.694452 (lr=2.1945e-06) (hash(x)=40842546)
|
| 273 |
+
2250 train 6.624052 (lr=2.1429e-06) (hash(x)=32055949)
|
| 274 |
+
2260 train 6.625170 (lr=2.0933e-06) (hash(x)=35806944)
|
| 275 |
+
2270 train 6.864573 (lr=2.0455e-06) (hash(x)=43553351)
|
| 276 |
+
2280 train 6.897438 (lr=1.9997e-06) (hash(x)=33250224)
|
| 277 |
+
2290 train 6.814785 (lr=1.9558e-06) (hash(x)=37657105)
|
| 278 |
+
2300 val loss 6.8601
|
| 279 |
+
2300 val perplexity 953.4459
|
| 280 |
+
2300 train 6.655499 (lr=1.9139e-06) (hash(x)=40443248)
|
| 281 |
+
2310 train 6.826280 (lr=1.8739e-06) (hash(x)=35689864)
|
| 282 |
+
2320 train 7.021714 (lr=1.8359e-06) (hash(x)=44241023)
|
| 283 |
+
2330 train 6.872594 (lr=1.7999e-06) (hash(x)=43375051)
|
| 284 |
+
2340 train 6.915829 (lr=1.7659e-06) (hash(x)=41936227)
|
| 285 |
+
2350 train 6.948425 (lr=1.7339e-06) (hash(x)=39892044)
|
| 286 |
+
2360 train 6.752195 (lr=1.7039e-06) (hash(x)=42188350)
|
| 287 |
+
2370 train 6.759420 (lr=1.6759e-06) (hash(x)=39758002)
|
| 288 |
+
2380 train 6.729066 (lr=1.6500e-06) (hash(x)=34237373)
|
| 289 |
+
2390 train 6.715007 (lr=1.6261e-06) (hash(x)=36323919)
|
| 290 |
+
2400 val loss 6.8532
|
| 291 |
+
2400 val perplexity 946.8890
|
| 292 |
+
2400 train 6.707891 (lr=1.6043e-06) (hash(x)=34074032)
|
| 293 |
+
2410 train 6.629775 (lr=1.5845e-06) (hash(x)=34145856)
|
| 294 |
+
2420 train 6.720111 (lr=1.5668e-06) (hash(x)=37905974)
|
| 295 |
+
2430 train 6.743615 (lr=1.5512e-06) (hash(x)=46570873)
|
| 296 |
+
2440 train 6.764393 (lr=1.5376e-06) (hash(x)=37393321)
|
| 297 |
+
2450 train 6.661545 (lr=1.5261e-06) (hash(x)=36905360)
|
| 298 |
+
2460 train 6.765852 (lr=1.5167e-06) (hash(x)=40719368)
|
| 299 |
+
2470 train 6.696073 (lr=1.5094e-06) (hash(x)=40499528)
|
| 300 |
+
2480 train 6.810740 (lr=1.5042e-06) (hash(x)=41998640)
|
| 301 |
+
2490 train 6.761917 (lr=1.5010e-06) (hash(x)=41290438)
|
| 302 |
+
2499 val loss 6.8515
|
| 303 |
+
2499 val perplexity 945.3420
|
4x_smaller_bs_half_lr_half_seq_len/model_02499.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3eb06466b0ff3a942193bc5258fda0a8e4dce50a9835ef93150cf826b42b2ed0
|
| 3 |
+
size 508831106
|
4x_smaller_bs_half_lr_half_seq_len/optimizer_02499.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5550341cd898e3dfd0a75907622f1270519bea0386c5f7efdab4f5a7b282dccf
|
| 3 |
+
size 992507142
|