Upload folder using huggingface_hub
Browse files
training-runs/byte-v15/tokenizer/meta.json
ADDED
|
@@ -0,0 +1,112 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"metrics": {
|
| 3 |
+
"token_f1": 97.37,
|
| 4 |
+
"prec": 97.06,
|
| 5 |
+
"rec": 97.69,
|
| 6 |
+
"byte_acc": 99.61,
|
| 7 |
+
"per_lang": {
|
| 8 |
+
"deu": {
|
| 9 |
+
"token_f1": 99.8,
|
| 10 |
+
"prec": 99.74,
|
| 11 |
+
"rec": 99.85,
|
| 12 |
+
"byte_acc": 99.96
|
| 13 |
+
},
|
| 14 |
+
"eng": {
|
| 15 |
+
"token_f1": 99.32,
|
| 16 |
+
"prec": 99.11,
|
| 17 |
+
"rec": 99.53,
|
| 18 |
+
"byte_acc": 99.83
|
| 19 |
+
},
|
| 20 |
+
"fra": {
|
| 21 |
+
"token_f1": 99.2,
|
| 22 |
+
"prec": 98.94,
|
| 23 |
+
"rec": 99.47,
|
| 24 |
+
"byte_acc": 99.82
|
| 25 |
+
},
|
| 26 |
+
"hin": {
|
| 27 |
+
"token_f1": 95.61,
|
| 28 |
+
"prec": 94.22,
|
| 29 |
+
"rec": 97.04,
|
| 30 |
+
"byte_acc": 99.39
|
| 31 |
+
},
|
| 32 |
+
"ita": {
|
| 33 |
+
"token_f1": 99.23,
|
| 34 |
+
"prec": 98.99,
|
| 35 |
+
"rec": 99.47,
|
| 36 |
+
"byte_acc": 99.8
|
| 37 |
+
},
|
| 38 |
+
"jpn": {
|
| 39 |
+
"token_f1": 94.51,
|
| 40 |
+
"prec": 93.51,
|
| 41 |
+
"rec": 95.53,
|
| 42 |
+
"byte_acc": 99.33
|
| 43 |
+
},
|
| 44 |
+
"kor": {
|
| 45 |
+
"token_f1": 93.22,
|
| 46 |
+
"prec": 93.46,
|
| 47 |
+
"rec": 92.99,
|
| 48 |
+
"byte_acc": 99.19
|
| 49 |
+
},
|
| 50 |
+
"por": {
|
| 51 |
+
"token_f1": 99.97,
|
| 52 |
+
"prec": 99.97,
|
| 53 |
+
"rec": 99.97,
|
| 54 |
+
"byte_acc": 99.99
|
| 55 |
+
},
|
| 56 |
+
"rus": {
|
| 57 |
+
"token_f1": 99.71,
|
| 58 |
+
"prec": 99.64,
|
| 59 |
+
"rec": 99.78,
|
| 60 |
+
"byte_acc": 99.95
|
| 61 |
+
},
|
| 62 |
+
"spa": {
|
| 63 |
+
"token_f1": 99.14,
|
| 64 |
+
"prec": 98.88,
|
| 65 |
+
"rec": 99.4,
|
| 66 |
+
"byte_acc": 99.76
|
| 67 |
+
},
|
| 68 |
+
"tha": {
|
| 69 |
+
"token_f1": 93.71,
|
| 70 |
+
"prec": 92.36,
|
| 71 |
+
"rec": 95.11,
|
| 72 |
+
"byte_acc": 99.46
|
| 73 |
+
},
|
| 74 |
+
"zho-hans": {
|
| 75 |
+
"token_f1": 94.08,
|
| 76 |
+
"prec": 94.78,
|
| 77 |
+
"rec": 93.39,
|
| 78 |
+
"byte_acc": 99.03
|
| 79 |
+
}
|
| 80 |
+
}
|
| 81 |
+
},
|
| 82 |
+
"step": 56000,
|
| 83 |
+
"config": {
|
| 84 |
+
"data_dir": "data/processed",
|
| 85 |
+
"out_dir": "output/tokenizer",
|
| 86 |
+
"hidden": 192,
|
| 87 |
+
"layers": 4,
|
| 88 |
+
"emb_dim": 96,
|
| 89 |
+
"lang_dropout": 0.15,
|
| 90 |
+
"max_bytes": 512,
|
| 91 |
+
"epochs": 2.0,
|
| 92 |
+
"batch_size": 128,
|
| 93 |
+
"lr": 0.002,
|
| 94 |
+
"workers": 8,
|
| 95 |
+
"log_every": 100,
|
| 96 |
+
"eval_every": 2000,
|
| 97 |
+
"train_limit": null,
|
| 98 |
+
"merge_inflection": true,
|
| 99 |
+
"use_prior": true,
|
| 100 |
+
"prior_mode": "concat",
|
| 101 |
+
"prior_dim": 8,
|
| 102 |
+
"langs": null,
|
| 103 |
+
"wordbank_langs": "kor,hin",
|
| 104 |
+
"wordbank_dir": "data/wordbanks",
|
| 105 |
+
"prior_sidecar": "data/processed/priors",
|
| 106 |
+
"prior_warmup_frac": 0.5,
|
| 107 |
+
"prior_dropout": 0.1,
|
| 108 |
+
"group_unknown": true,
|
| 109 |
+
"smoke": false,
|
| 110 |
+
"wandb": false
|
| 111 |
+
}
|
| 112 |
+
}
|
training-runs/byte-v15/tokenizer/tokenizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:caaf3bb22fa002f603ccafcd82c19796c8647b096067962886f92976d4d5ccd3
|
| 3 |
+
size 3996297
|