Upload folder using huggingface_hub
Browse files- config.json +95 -0
- model_best.pth +3 -0
config.json
ADDED
|
@@ -0,0 +1,95 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"experiment": "upos",
|
| 3 |
+
"name": "upos_xlmr_tr_1",
|
| 4 |
+
"n_gpu": 1,
|
| 5 |
+
"model": {
|
| 6 |
+
"type": "MultiParser",
|
| 7 |
+
"args": {
|
| 8 |
+
"embeddings_processor": {
|
| 9 |
+
"type": "XLMRobertaWrapper",
|
| 10 |
+
"args": {
|
| 11 |
+
"model_path": "data/pretrained_embeddings/xlm-roberta-base",
|
| 12 |
+
"fine_tune": true,
|
| 13 |
+
"hidden_dropout": 0.2,
|
| 14 |
+
"attn_dropout": 0.2,
|
| 15 |
+
"output_dropout": 0.5,
|
| 16 |
+
"scalar_mix_layer_dropout": 0.1,
|
| 17 |
+
"token_mask_prob": 0.15
|
| 18 |
+
}
|
| 19 |
+
},
|
| 20 |
+
"outputs": {
|
| 21 |
+
"upos": {
|
| 22 |
+
"type": "SequenceTagger",
|
| 23 |
+
"args": {
|
| 24 |
+
"hidden_size": 0,
|
| 25 |
+
"input_dropout": 0.2,
|
| 26 |
+
"vocab": {
|
| 27 |
+
"type": "BasicVocab",
|
| 28 |
+
"args": {
|
| 29 |
+
"vocab_filename": "data/corpora/ota_boun/vocab/upos.vocab"
|
| 30 |
+
}
|
| 31 |
+
}
|
| 32 |
+
}
|
| 33 |
+
}
|
| 34 |
+
},
|
| 35 |
+
"post_processors": []
|
| 36 |
+
}
|
| 37 |
+
},
|
| 38 |
+
"data_loaders": {
|
| 39 |
+
"type": "BucketedCONLLLoader",
|
| 40 |
+
"args": {
|
| 41 |
+
"annotation_layers": {
|
| 42 |
+
"upos": {
|
| 43 |
+
"type": "TagSequence",
|
| 44 |
+
"source_column": 3,
|
| 45 |
+
"args": {
|
| 46 |
+
"ignore_root": true
|
| 47 |
+
}
|
| 48 |
+
}
|
| 49 |
+
},
|
| 50 |
+
"batch_size": 32,
|
| 51 |
+
"bucket_size": 512,
|
| 52 |
+
"max_tokens_per_batch": 156800,
|
| 53 |
+
"num_workers": 0
|
| 54 |
+
},
|
| 55 |
+
"paths": {
|
| 56 |
+
"train": "data/corpora/ota_boun/tr_boun-ud-train.conllu",
|
| 57 |
+
"dev": "data/corpora/ota_boun/tr_boun-ud-dev.conllu",
|
| 58 |
+
"test_ota": "data/corpora/ota_boun/ota_boun-test-2026.conllu",
|
| 59 |
+
"test_tr": "data/corpora/ota_boun/tr_boun-ud-test.conllu"
|
| 60 |
+
}
|
| 61 |
+
},
|
| 62 |
+
"trainer": {
|
| 63 |
+
"min_epochs": 15,
|
| 64 |
+
"max_epochs": 300,
|
| 65 |
+
"early_stop": 15,
|
| 66 |
+
"save_dir": "/content/drive/MyDrive/hf_models/saved_models",
|
| 67 |
+
"save_period": 20,
|
| 68 |
+
"verbosity": 2,
|
| 69 |
+
"validation_criterion": {
|
| 70 |
+
"metrics": {
|
| 71 |
+
"upos": "fscore"
|
| 72 |
+
},
|
| 73 |
+
"weighting": "multiplicative"
|
| 74 |
+
},
|
| 75 |
+
"optimizer": {
|
| 76 |
+
"type": "AdamW",
|
| 77 |
+
"args": {
|
| 78 |
+
"lr": 4e-05,
|
| 79 |
+
"weight_decay": 0.0
|
| 80 |
+
}
|
| 81 |
+
},
|
| 82 |
+
"lr_scheduler": {
|
| 83 |
+
"type": "LambdaLR",
|
| 84 |
+
"args": {
|
| 85 |
+
"lr_lambda": "SqrtSchedule(400)"
|
| 86 |
+
}
|
| 87 |
+
},
|
| 88 |
+
"loss": {
|
| 89 |
+
"type": "CrossEntropyLoss",
|
| 90 |
+
"args": {
|
| 91 |
+
"ignore_index": -1
|
| 92 |
+
}
|
| 93 |
+
}
|
| 94 |
+
}
|
| 95 |
+
}
|
model_best.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:79367640a443412b1a31f8105d16a46a91999c102dc0cabb8b6969a93eb5e7d4
|
| 3 |
+
size 3332222462
|