Upload 3 files
Browse files
baseline_transformer_tinystories.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6102d768719877d214395643421f3765ba5a9a384db673da8e05e12ce39cdc5a
|
| 3 |
+
size 139852941
|
baseline_transformer_tinystories_tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
baseline_transformer_tinystories_training_config.json
ADDED
|
@@ -0,0 +1,148 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"run_name": "baseline_transformer_tinystories",
|
| 3 |
+
"checkpoint_tag": "final",
|
| 4 |
+
"epoch_value": 2.0,
|
| 5 |
+
"dataset": "roneneldan/TinyStories",
|
| 6 |
+
"num_train_texts": 100000,
|
| 7 |
+
"eval_texts": 2000,
|
| 8 |
+
"max_seq_len": 128,
|
| 9 |
+
"max_vocab_size": 48000,
|
| 10 |
+
"actual_vocab_size": 18604,
|
| 11 |
+
"param_count": 34949824,
|
| 12 |
+
"model_config": {
|
| 13 |
+
"d_model": 448,
|
| 14 |
+
"n_layers": 11,
|
| 15 |
+
"n_heads": 4,
|
| 16 |
+
"d_ff": 1792
|
| 17 |
+
},
|
| 18 |
+
"history": [
|
| 19 |
+
{
|
| 20 |
+
"epoch": 1,
|
| 21 |
+
"loss": 2.7925477427443095,
|
| 22 |
+
"ce": 2.7925477427443095,
|
| 23 |
+
"ppl": 16.322552536754753,
|
| 24 |
+
"seconds": 6709.080967903137,
|
| 25 |
+
"global_step": 13778
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 2,
|
| 29 |
+
"loss": 2.145891742479183,
|
| 30 |
+
"ce": 2.145891742479183,
|
| 31 |
+
"ppl": 8.54966193566955,
|
| 32 |
+
"seconds": 6647.170892953873,
|
| 33 |
+
"global_step": 27556
|
| 34 |
+
}
|
| 35 |
+
],
|
| 36 |
+
"eval_history": [
|
| 37 |
+
{
|
| 38 |
+
"eval_ce": 3.019229908338063,
|
| 39 |
+
"eval_ppl": 20.475517585854988,
|
| 40 |
+
"eval_tokens": 418011,
|
| 41 |
+
"eval_batches": 4396,
|
| 42 |
+
"trigger_progress": 0.1,
|
| 43 |
+
"epoch_value": 0.2000027218661114,
|
| 44 |
+
"global_step": 2755,
|
| 45 |
+
"train_ce_running": 3.726730931783734,
|
| 46 |
+
"train_ppl_running": 41.54307878205579
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"eval_ce": 2.6282812372172235,
|
| 50 |
+
"eval_ppl": 13.84994466359723,
|
| 51 |
+
"eval_tokens": 418011,
|
| 52 |
+
"eval_batches": 4396,
|
| 53 |
+
"trigger_progress": 0.2,
|
| 54 |
+
"epoch_value": 0.4000009072887038,
|
| 55 |
+
"global_step": 5510,
|
| 56 |
+
"train_ce_running": 3.298934572113122,
|
| 57 |
+
"train_ppl_running": 27.083767741875523
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"eval_ce": 2.4289824040710375,
|
| 61 |
+
"eval_ppl": 11.34732920691451,
|
| 62 |
+
"eval_tokens": 418011,
|
| 63 |
+
"eval_batches": 4396,
|
| 64 |
+
"trigger_progress": 0.3,
|
| 65 |
+
"epoch_value": 0.6000036291548152,
|
| 66 |
+
"global_step": 8266,
|
| 67 |
+
"train_ce_running": 3.0639317816293103,
|
| 68 |
+
"train_ppl_running": 21.411577526098043
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"eval_ce": 2.3028478895985414,
|
| 72 |
+
"eval_ppl": 10.002628311385484,
|
| 73 |
+
"eval_tokens": 418011,
|
| 74 |
+
"eval_batches": 4396,
|
| 75 |
+
"trigger_progress": 0.4,
|
| 76 |
+
"epoch_value": 0.8000018145774076,
|
| 77 |
+
"global_step": 11021,
|
| 78 |
+
"train_ce_running": 2.9075046430774236,
|
| 79 |
+
"train_ppl_running": 18.311048907242007
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"eval_ce": 2.2137269049744646,
|
| 83 |
+
"eval_ppl": 9.14975318836517,
|
| 84 |
+
"eval_tokens": 418011,
|
| 85 |
+
"eval_batches": 4396,
|
| 86 |
+
"trigger_progress": 0.5,
|
| 87 |
+
"epoch_value": 1.0,
|
| 88 |
+
"global_step": 13777,
|
| 89 |
+
"train_ce_running": 2.7925477427443095,
|
| 90 |
+
"train_ppl_running": 16.322552536754753
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"eval_ce": 2.1519488045059103,
|
| 94 |
+
"eval_ppl": 8.601604920454877,
|
| 95 |
+
"eval_tokens": 418011,
|
| 96 |
+
"eval_batches": 4396,
|
| 97 |
+
"trigger_progress": 0.6,
|
| 98 |
+
"epoch_value": 1.2000027218661113,
|
| 99 |
+
"global_step": 16533,
|
| 100 |
+
"train_ce_running": 2.2201592317098897,
|
| 101 |
+
"train_ppl_running": 9.20879708165047
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"eval_ce": 2.1023585491979833,
|
| 105 |
+
"eval_ppl": 8.185452957131274,
|
| 106 |
+
"eval_tokens": 418011,
|
| 107 |
+
"eval_batches": 4396,
|
| 108 |
+
"trigger_progress": 0.7,
|
| 109 |
+
"epoch_value": 1.4000009072887039,
|
| 110 |
+
"global_step": 19288,
|
| 111 |
+
"train_ce_running": 2.199022379118235,
|
| 112 |
+
"train_ppl_running": 9.01619476916866
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"eval_ce": 2.0675348904853807,
|
| 116 |
+
"eval_ppl": 7.905311619850248,
|
| 117 |
+
"eval_tokens": 418011,
|
| 118 |
+
"eval_batches": 4396,
|
| 119 |
+
"trigger_progress": 0.8,
|
| 120 |
+
"epoch_value": 1.6000036291548152,
|
| 121 |
+
"global_step": 22044,
|
| 122 |
+
"train_ce_running": 2.178444489460884,
|
| 123 |
+
"train_ppl_running": 8.83255643287514
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"eval_ce": 2.045298832631953,
|
| 127 |
+
"eval_ppl": 7.731468607843494,
|
| 128 |
+
"eval_tokens": 418011,
|
| 129 |
+
"eval_batches": 4396,
|
| 130 |
+
"trigger_progress": 0.9,
|
| 131 |
+
"epoch_value": 1.8000018145774077,
|
| 132 |
+
"global_step": 24799,
|
| 133 |
+
"train_ce_running": 2.160078341528155,
|
| 134 |
+
"train_ppl_running": 8.671816995248255
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"eval_ce": 2.0341344709381657,
|
| 138 |
+
"eval_ppl": 7.645631744354986,
|
| 139 |
+
"eval_tokens": 418011,
|
| 140 |
+
"eval_batches": 4396,
|
| 141 |
+
"trigger_progress": 1.0,
|
| 142 |
+
"epoch_value": 2.0,
|
| 143 |
+
"global_step": 27555,
|
| 144 |
+
"train_ce_running": 2.145891742479183,
|
| 145 |
+
"train_ppl_running": 8.54966193566955
|
| 146 |
+
}
|
| 147 |
+
]
|
| 148 |
+
}
|