Ethan Sim
commited on
Commit
•
2e7b5b1
1
Parent(s):
b0adf36
stage best wang model
Browse files- checkpoint-256000/config.json +37 -0
- checkpoint-256000/generation_config.json +10 -0
- checkpoint-256000/optimizer.pt +3 -0
- checkpoint-256000/pytorch_model.bin +3 -0
- checkpoint-256000/rng_state.pth +3 -0
- checkpoint-256000/scaler.pt +3 -0
- checkpoint-256000/scheduler.pt +3 -0
- checkpoint-256000/sentencepiece.bpe.model +3 -0
- checkpoint-256000/special_tokens_map.json +109 -0
- checkpoint-256000/tokenizer_config.json +118 -0
- checkpoint-256000/trainer_state.json +272 -0
- checkpoint-256000/training_args.bin +3 -0
- checkpoint-256000/vocab.json +0 -0
- config.json +37 -0
- generation_config.json +10 -0
- pytorch_model.bin +3 -0
- sentencepiece.bpe.model +3 -0
- special_tokens_map.json +109 -0
- tokenizer_config.json +118 -0
- training_args.bin +3 -0
- vocab.json +0 -0
checkpoint-256000/config.json
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "facebook/m2m100_418M",
|
| 3 |
+
"activation_dropout": 0.0,
|
| 4 |
+
"activation_function": "relu",
|
| 5 |
+
"architectures": [
|
| 6 |
+
"M2M100ForConditionalGeneration"
|
| 7 |
+
],
|
| 8 |
+
"attention_dropout": 0.1,
|
| 9 |
+
"bos_token_id": 0,
|
| 10 |
+
"d_model": 1024,
|
| 11 |
+
"decoder_attention_heads": 16,
|
| 12 |
+
"decoder_ffn_dim": 4096,
|
| 13 |
+
"decoder_layerdrop": 0.05,
|
| 14 |
+
"decoder_layers": 12,
|
| 15 |
+
"decoder_start_token_id": 2,
|
| 16 |
+
"dropout": 0.1,
|
| 17 |
+
"early_stopping": true,
|
| 18 |
+
"encoder_attention_heads": 16,
|
| 19 |
+
"encoder_ffn_dim": 4096,
|
| 20 |
+
"encoder_layerdrop": 0.05,
|
| 21 |
+
"encoder_layers": 12,
|
| 22 |
+
"eos_token_id": 2,
|
| 23 |
+
"gradient_checkpointing": false,
|
| 24 |
+
"init_std": 0.02,
|
| 25 |
+
"is_encoder_decoder": true,
|
| 26 |
+
"max_length": 200,
|
| 27 |
+
"max_position_embeddings": 1024,
|
| 28 |
+
"model_type": "m2m_100",
|
| 29 |
+
"num_beams": 5,
|
| 30 |
+
"num_hidden_layers": 12,
|
| 31 |
+
"pad_token_id": 1,
|
| 32 |
+
"scale_embedding": true,
|
| 33 |
+
"torch_dtype": "float32",
|
| 34 |
+
"transformers_version": "4.28.0",
|
| 35 |
+
"use_cache": true,
|
| 36 |
+
"vocab_size": 128112
|
| 37 |
+
}
|
checkpoint-256000/generation_config.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 0,
|
| 3 |
+
"decoder_start_token_id": 2,
|
| 4 |
+
"early_stopping": true,
|
| 5 |
+
"eos_token_id": 2,
|
| 6 |
+
"max_length": 200,
|
| 7 |
+
"num_beams": 5,
|
| 8 |
+
"pad_token_id": 1,
|
| 9 |
+
"transformers_version": "4.28.0"
|
| 10 |
+
}
|
checkpoint-256000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:38a1322634b2d217675b81412e30f3374b08abf7623ae73d2e96d9efb795e2fe
|
| 3 |
+
size 3871544599
|
checkpoint-256000/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c353a0da2bdb44caf3bc774a1f09200622a89d1206338b7f9ac554bd3cd9171d
|
| 3 |
+
size 1944201353
|
checkpoint-256000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f8aaf9520cf4149c556ce815dc4ff12f8d7a864f94b792e1a28c43d73bc3e021
|
| 3 |
+
size 14511
|
checkpoint-256000/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:47acc835a3f0fabec0bed149f23e92d29a899f1a62e0cd51b616df2f6058e03f
|
| 3 |
+
size 557
|
checkpoint-256000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:09db2bddec8d85cd09c3ced5c24082ea85f7fc1208e00ae381ead78d58fc37c3
|
| 3 |
+
size 627
|
checkpoint-256000/sentencepiece.bpe.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d8f7c76ed2a5e0822be39f0a4f95a55eb19c78f4593ce609e2edbc2aea4d380a
|
| 3 |
+
size 2423393
|
checkpoint-256000/special_tokens_map.json
ADDED
|
@@ -0,0 +1,109 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"__af__",
|
| 4 |
+
"__am__",
|
| 5 |
+
"__ar__",
|
| 6 |
+
"__ast__",
|
| 7 |
+
"__az__",
|
| 8 |
+
"__ba__",
|
| 9 |
+
"__be__",
|
| 10 |
+
"__bg__",
|
| 11 |
+
"__bn__",
|
| 12 |
+
"__br__",
|
| 13 |
+
"__bs__",
|
| 14 |
+
"__ca__",
|
| 15 |
+
"__ceb__",
|
| 16 |
+
"__cs__",
|
| 17 |
+
"__cy__",
|
| 18 |
+
"__da__",
|
| 19 |
+
"__de__",
|
| 20 |
+
"__el__",
|
| 21 |
+
"__en__",
|
| 22 |
+
"__es__",
|
| 23 |
+
"__et__",
|
| 24 |
+
"__fa__",
|
| 25 |
+
"__ff__",
|
| 26 |
+
"__fi__",
|
| 27 |
+
"__fr__",
|
| 28 |
+
"__fy__",
|
| 29 |
+
"__ga__",
|
| 30 |
+
"__gd__",
|
| 31 |
+
"__gl__",
|
| 32 |
+
"__gu__",
|
| 33 |
+
"__ha__",
|
| 34 |
+
"__he__",
|
| 35 |
+
"__hi__",
|
| 36 |
+
"__hr__",
|
| 37 |
+
"__ht__",
|
| 38 |
+
"__hu__",
|
| 39 |
+
"__hy__",
|
| 40 |
+
"__id__",
|
| 41 |
+
"__ig__",
|
| 42 |
+
"__ilo__",
|
| 43 |
+
"__is__",
|
| 44 |
+
"__it__",
|
| 45 |
+
"__ja__",
|
| 46 |
+
"__jv__",
|
| 47 |
+
"__ka__",
|
| 48 |
+
"__kk__",
|
| 49 |
+
"__km__",
|
| 50 |
+
"__kn__",
|
| 51 |
+
"__ko__",
|
| 52 |
+
"__lb__",
|
| 53 |
+
"__lg__",
|
| 54 |
+
"__ln__",
|
| 55 |
+
"__lo__",
|
| 56 |
+
"__lt__",
|
| 57 |
+
"__lv__",
|
| 58 |
+
"__mg__",
|
| 59 |
+
"__mk__",
|
| 60 |
+
"__ml__",
|
| 61 |
+
"__mn__",
|
| 62 |
+
"__mr__",
|
| 63 |
+
"__ms__",
|
| 64 |
+
"__my__",
|
| 65 |
+
"__ne__",
|
| 66 |
+
"__nl__",
|
| 67 |
+
"__no__",
|
| 68 |
+
"__ns__",
|
| 69 |
+
"__oc__",
|
| 70 |
+
"__or__",
|
| 71 |
+
"__pa__",
|
| 72 |
+
"__pl__",
|
| 73 |
+
"__ps__",
|
| 74 |
+
"__pt__",
|
| 75 |
+
"__ro__",
|
| 76 |
+
"__ru__",
|
| 77 |
+
"__sd__",
|
| 78 |
+
"__si__",
|
| 79 |
+
"__sk__",
|
| 80 |
+
"__sl__",
|
| 81 |
+
"__so__",
|
| 82 |
+
"__sq__",
|
| 83 |
+
"__sr__",
|
| 84 |
+
"__ss__",
|
| 85 |
+
"__su__",
|
| 86 |
+
"__sv__",
|
| 87 |
+
"__sw__",
|
| 88 |
+
"__ta__",
|
| 89 |
+
"__th__",
|
| 90 |
+
"__tl__",
|
| 91 |
+
"__tn__",
|
| 92 |
+
"__tr__",
|
| 93 |
+
"__uk__",
|
| 94 |
+
"__ur__",
|
| 95 |
+
"__uz__",
|
| 96 |
+
"__vi__",
|
| 97 |
+
"__wo__",
|
| 98 |
+
"__xh__",
|
| 99 |
+
"__yi__",
|
| 100 |
+
"__yo__",
|
| 101 |
+
"__zh__",
|
| 102 |
+
"__zu__"
|
| 103 |
+
],
|
| 104 |
+
"bos_token": "<s>",
|
| 105 |
+
"eos_token": "</s>",
|
| 106 |
+
"pad_token": "<pad>",
|
| 107 |
+
"sep_token": "</s>",
|
| 108 |
+
"unk_token": "<unk>"
|
| 109 |
+
}
|
checkpoint-256000/tokenizer_config.json
ADDED
|
@@ -0,0 +1,118 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"__af__",
|
| 4 |
+
"__am__",
|
| 5 |
+
"__ar__",
|
| 6 |
+
"__ast__",
|
| 7 |
+
"__az__",
|
| 8 |
+
"__ba__",
|
| 9 |
+
"__be__",
|
| 10 |
+
"__bg__",
|
| 11 |
+
"__bn__",
|
| 12 |
+
"__br__",
|
| 13 |
+
"__bs__",
|
| 14 |
+
"__ca__",
|
| 15 |
+
"__ceb__",
|
| 16 |
+
"__cs__",
|
| 17 |
+
"__cy__",
|
| 18 |
+
"__da__",
|
| 19 |
+
"__de__",
|
| 20 |
+
"__el__",
|
| 21 |
+
"__en__",
|
| 22 |
+
"__es__",
|
| 23 |
+
"__et__",
|
| 24 |
+
"__fa__",
|
| 25 |
+
"__ff__",
|
| 26 |
+
"__fi__",
|
| 27 |
+
"__fr__",
|
| 28 |
+
"__fy__",
|
| 29 |
+
"__ga__",
|
| 30 |
+
"__gd__",
|
| 31 |
+
"__gl__",
|
| 32 |
+
"__gu__",
|
| 33 |
+
"__ha__",
|
| 34 |
+
"__he__",
|
| 35 |
+
"__hi__",
|
| 36 |
+
"__hr__",
|
| 37 |
+
"__ht__",
|
| 38 |
+
"__hu__",
|
| 39 |
+
"__hy__",
|
| 40 |
+
"__id__",
|
| 41 |
+
"__ig__",
|
| 42 |
+
"__ilo__",
|
| 43 |
+
"__is__",
|
| 44 |
+
"__it__",
|
| 45 |
+
"__ja__",
|
| 46 |
+
"__jv__",
|
| 47 |
+
"__ka__",
|
| 48 |
+
"__kk__",
|
| 49 |
+
"__km__",
|
| 50 |
+
"__kn__",
|
| 51 |
+
"__ko__",
|
| 52 |
+
"__lb__",
|
| 53 |
+
"__lg__",
|
| 54 |
+
"__ln__",
|
| 55 |
+
"__lo__",
|
| 56 |
+
"__lt__",
|
| 57 |
+
"__lv__",
|
| 58 |
+
"__mg__",
|
| 59 |
+
"__mk__",
|
| 60 |
+
"__ml__",
|
| 61 |
+
"__mn__",
|
| 62 |
+
"__mr__",
|
| 63 |
+
"__ms__",
|
| 64 |
+
"__my__",
|
| 65 |
+
"__ne__",
|
| 66 |
+
"__nl__",
|
| 67 |
+
"__no__",
|
| 68 |
+
"__ns__",
|
| 69 |
+
"__oc__",
|
| 70 |
+
"__or__",
|
| 71 |
+
"__pa__",
|
| 72 |
+
"__pl__",
|
| 73 |
+
"__ps__",
|
| 74 |
+
"__pt__",
|
| 75 |
+
"__ro__",
|
| 76 |
+
"__ru__",
|
| 77 |
+
"__sd__",
|
| 78 |
+
"__si__",
|
| 79 |
+
"__sk__",
|
| 80 |
+
"__sl__",
|
| 81 |
+
"__so__",
|
| 82 |
+
"__sq__",
|
| 83 |
+
"__sr__",
|
| 84 |
+
"__ss__",
|
| 85 |
+
"__su__",
|
| 86 |
+
"__sv__",
|
| 87 |
+
"__sw__",
|
| 88 |
+
"__ta__",
|
| 89 |
+
"__th__",
|
| 90 |
+
"__tl__",
|
| 91 |
+
"__tn__",
|
| 92 |
+
"__tr__",
|
| 93 |
+
"__uk__",
|
| 94 |
+
"__ur__",
|
| 95 |
+
"__uz__",
|
| 96 |
+
"__vi__",
|
| 97 |
+
"__wo__",
|
| 98 |
+
"__xh__",
|
| 99 |
+
"__yi__",
|
| 100 |
+
"__yo__",
|
| 101 |
+
"__zh__",
|
| 102 |
+
"__zu__"
|
| 103 |
+
],
|
| 104 |
+
"bos_token": "<s>",
|
| 105 |
+
"clean_up_tokenization_spaces": true,
|
| 106 |
+
"eos_token": "</s>",
|
| 107 |
+
"language_codes": "m2m100",
|
| 108 |
+
"model_max_length": 1024,
|
| 109 |
+
"num_madeup_words": 8,
|
| 110 |
+
"pad_token": "<pad>",
|
| 111 |
+
"sep_token": "</s>",
|
| 112 |
+
"sp_model_kwargs": {},
|
| 113 |
+
"src_lang": "en",
|
| 114 |
+
"tgt_lang": "fr",
|
| 115 |
+
"tokenizer_class": "M2M100Tokenizer",
|
| 116 |
+
"tokenizer_file": null,
|
| 117 |
+
"unk_token": "<unk>"
|
| 118 |
+
}
|
checkpoint-256000/trainer_state.json
ADDED
|
@@ -0,0 +1,272 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 40.6424,
|
| 3 |
+
"best_model_checkpoint": "M2M100_enfr_FT_wang_2022/checkpoint-256000",
|
| 4 |
+
"epoch": 3.130158341994253,
|
| 5 |
+
"global_step": 256000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.2,
|
| 12 |
+
"learning_rate": 1.975551751543682e-05,
|
| 13 |
+
"loss": 0.2244,
|
| 14 |
+
"step": 16000
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.2,
|
| 18 |
+
"eval_bleu": 35.6567,
|
| 19 |
+
"eval_gen_len": 44.8111,
|
| 20 |
+
"eval_loss": 0.13247372210025787,
|
| 21 |
+
"eval_runtime": 433.7309,
|
| 22 |
+
"eval_samples_per_second": 2.405,
|
| 23 |
+
"eval_steps_per_second": 0.302,
|
| 24 |
+
"step": 16000
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.39,
|
| 28 |
+
"learning_rate": 1.951108088280247e-05,
|
| 29 |
+
"loss": 0.1522,
|
| 30 |
+
"step": 32000
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.39,
|
| 34 |
+
"eval_bleu": 36.512,
|
| 35 |
+
"eval_gen_len": 44.4851,
|
| 36 |
+
"eval_loss": 0.1262633502483368,
|
| 37 |
+
"eval_runtime": 417.8212,
|
| 38 |
+
"eval_samples_per_second": 2.496,
|
| 39 |
+
"eval_steps_per_second": 0.314,
|
| 40 |
+
"step": 32000
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"epoch": 0.59,
|
| 44 |
+
"learning_rate": 1.9266613682215567e-05,
|
| 45 |
+
"loss": 0.1435,
|
| 46 |
+
"step": 48000
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.59,
|
| 50 |
+
"eval_bleu": 37.957,
|
| 51 |
+
"eval_gen_len": 44.5302,
|
| 52 |
+
"eval_loss": 0.12126067280769348,
|
| 53 |
+
"eval_runtime": 414.4742,
|
| 54 |
+
"eval_samples_per_second": 2.516,
|
| 55 |
+
"eval_steps_per_second": 0.316,
|
| 56 |
+
"step": 48000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.78,
|
| 60 |
+
"learning_rate": 1.9022177049581222e-05,
|
| 61 |
+
"loss": 0.1384,
|
| 62 |
+
"step": 64000
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.78,
|
| 66 |
+
"eval_bleu": 38.0569,
|
| 67 |
+
"eval_gen_len": 44.5034,
|
| 68 |
+
"eval_loss": 0.1191168949007988,
|
| 69 |
+
"eval_runtime": 415.8111,
|
| 70 |
+
"eval_samples_per_second": 2.508,
|
| 71 |
+
"eval_steps_per_second": 0.315,
|
| 72 |
+
"step": 64000
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.98,
|
| 76 |
+
"learning_rate": 1.8777709848994316e-05,
|
| 77 |
+
"loss": 0.1345,
|
| 78 |
+
"step": 80000
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 0.98,
|
| 82 |
+
"eval_bleu": 38.4966,
|
| 83 |
+
"eval_gen_len": 44.8821,
|
| 84 |
+
"eval_loss": 0.11713190376758575,
|
| 85 |
+
"eval_runtime": 423.2079,
|
| 86 |
+
"eval_samples_per_second": 2.465,
|
| 87 |
+
"eval_steps_per_second": 0.31,
|
| 88 |
+
"step": 80000
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 1.17,
|
| 92 |
+
"learning_rate": 1.853325793238369e-05,
|
| 93 |
+
"loss": 0.1213,
|
| 94 |
+
"step": 96000
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 1.17,
|
| 98 |
+
"eval_bleu": 39.0368,
|
| 99 |
+
"eval_gen_len": 44.6012,
|
| 100 |
+
"eval_loss": 0.1156671866774559,
|
| 101 |
+
"eval_runtime": 417.9213,
|
| 102 |
+
"eval_samples_per_second": 2.496,
|
| 103 |
+
"eval_steps_per_second": 0.313,
|
| 104 |
+
"step": 96000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 1.37,
|
| 108 |
+
"learning_rate": 1.8288806015773065e-05,
|
| 109 |
+
"loss": 0.1199,
|
| 110 |
+
"step": 112000
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 1.37,
|
| 114 |
+
"eval_bleu": 39.6906,
|
| 115 |
+
"eval_gen_len": 44.8178,
|
| 116 |
+
"eval_loss": 0.11390843987464905,
|
| 117 |
+
"eval_runtime": 418.2954,
|
| 118 |
+
"eval_samples_per_second": 2.493,
|
| 119 |
+
"eval_steps_per_second": 0.313,
|
| 120 |
+
"step": 112000
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 1.57,
|
| 124 |
+
"learning_rate": 1.804436938313872e-05,
|
| 125 |
+
"loss": 0.1195,
|
| 126 |
+
"step": 128000
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 1.57,
|
| 130 |
+
"eval_bleu": 39.6284,
|
| 131 |
+
"eval_gen_len": 44.8552,
|
| 132 |
+
"eval_loss": 0.11290750652551651,
|
| 133 |
+
"eval_runtime": 422.1564,
|
| 134 |
+
"eval_samples_per_second": 2.471,
|
| 135 |
+
"eval_steps_per_second": 0.31,
|
| 136 |
+
"step": 128000
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 1.76,
|
| 140 |
+
"learning_rate": 1.7799902182551813e-05,
|
| 141 |
+
"loss": 0.1185,
|
| 142 |
+
"step": 144000
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 1.76,
|
| 146 |
+
"eval_bleu": 39.0079,
|
| 147 |
+
"eval_gen_len": 44.5618,
|
| 148 |
+
"eval_loss": 0.11247587949037552,
|
| 149 |
+
"eval_runtime": 408.9786,
|
| 150 |
+
"eval_samples_per_second": 2.55,
|
| 151 |
+
"eval_steps_per_second": 0.32,
|
| 152 |
+
"step": 144000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 1.96,
|
| 156 |
+
"learning_rate": 1.7555465549917468e-05,
|
| 157 |
+
"loss": 0.1175,
|
| 158 |
+
"step": 160000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.96,
|
| 162 |
+
"eval_bleu": 39.2898,
|
| 163 |
+
"eval_gen_len": 44.745,
|
| 164 |
+
"eval_loss": 0.1103997528553009,
|
| 165 |
+
"eval_runtime": 413.1156,
|
| 166 |
+
"eval_samples_per_second": 2.525,
|
| 167 |
+
"eval_steps_per_second": 0.317,
|
| 168 |
+
"step": 160000
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 2.15,
|
| 172 |
+
"learning_rate": 1.7310998349330562e-05,
|
| 173 |
+
"loss": 0.1062,
|
| 174 |
+
"step": 176000
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 2.15,
|
| 178 |
+
"eval_bleu": 39.8593,
|
| 179 |
+
"eval_gen_len": 45.1151,
|
| 180 |
+
"eval_loss": 0.1111496165394783,
|
| 181 |
+
"eval_runtime": 415.6075,
|
| 182 |
+
"eval_samples_per_second": 2.51,
|
| 183 |
+
"eval_steps_per_second": 0.315,
|
| 184 |
+
"step": 176000
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 2.35,
|
| 188 |
+
"learning_rate": 1.7066546432719936e-05,
|
| 189 |
+
"loss": 0.1047,
|
| 190 |
+
"step": 192000
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 2.35,
|
| 194 |
+
"eval_bleu": 39.913,
|
| 195 |
+
"eval_gen_len": 44.8102,
|
| 196 |
+
"eval_loss": 0.11086419969797134,
|
| 197 |
+
"eval_runtime": 418.8388,
|
| 198 |
+
"eval_samples_per_second": 2.49,
|
| 199 |
+
"eval_steps_per_second": 0.313,
|
| 200 |
+
"step": 192000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 2.54,
|
| 204 |
+
"learning_rate": 1.682210980008559e-05,
|
| 205 |
+
"loss": 0.1055,
|
| 206 |
+
"step": 208000
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 2.54,
|
| 210 |
+
"eval_bleu": 40.2278,
|
| 211 |
+
"eval_gen_len": 45.2848,
|
| 212 |
+
"eval_loss": 0.11025020480155945,
|
| 213 |
+
"eval_runtime": 421.9598,
|
| 214 |
+
"eval_samples_per_second": 2.472,
|
| 215 |
+
"eval_steps_per_second": 0.31,
|
| 216 |
+
"step": 208000
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 2.74,
|
| 220 |
+
"learning_rate": 1.6577657883474966e-05,
|
| 221 |
+
"loss": 0.1059,
|
| 222 |
+
"step": 224000
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 2.74,
|
| 226 |
+
"eval_bleu": 40.2198,
|
| 227 |
+
"eval_gen_len": 45.0719,
|
| 228 |
+
"eval_loss": 0.10949720442295074,
|
| 229 |
+
"eval_runtime": 416.4666,
|
| 230 |
+
"eval_samples_per_second": 2.504,
|
| 231 |
+
"eval_steps_per_second": 0.315,
|
| 232 |
+
"step": 224000
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 2.93,
|
| 236 |
+
"learning_rate": 1.633320596686434e-05,
|
| 237 |
+
"loss": 0.106,
|
| 238 |
+
"step": 240000
|
| 239 |
+
},
|
| 240 |
+
{
|
| 241 |
+
"epoch": 2.93,
|
| 242 |
+
"eval_bleu": 39.8973,
|
| 243 |
+
"eval_gen_len": 44.954,
|
| 244 |
+
"eval_loss": 0.10881481319665909,
|
| 245 |
+
"eval_runtime": 421.0124,
|
| 246 |
+
"eval_samples_per_second": 2.477,
|
| 247 |
+
"eval_steps_per_second": 0.311,
|
| 248 |
+
"step": 240000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 3.13,
|
| 252 |
+
"learning_rate": 1.6088769334229995e-05,
|
| 253 |
+
"loss": 0.0971,
|
| 254 |
+
"step": 256000
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 3.13,
|
| 258 |
+
"eval_bleu": 40.6424,
|
| 259 |
+
"eval_gen_len": 44.9732,
|
| 260 |
+
"eval_loss": 0.11019956320524216,
|
| 261 |
+
"eval_runtime": 418.4973,
|
| 262 |
+
"eval_samples_per_second": 2.492,
|
| 263 |
+
"eval_steps_per_second": 0.313,
|
| 264 |
+
"step": 256000
|
| 265 |
+
}
|
| 266 |
+
],
|
| 267 |
+
"max_steps": 1308560,
|
| 268 |
+
"num_train_epochs": 16,
|
| 269 |
+
"total_flos": 8.283927408492872e+17,
|
| 270 |
+
"trial_name": null,
|
| 271 |
+
"trial_params": null
|
| 272 |
+
}
|
checkpoint-256000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:98438687991f47832564d8ab84087cb1c5a15d8246397eca599954545732c6e4
|
| 3 |
+
size 3771
|
checkpoint-256000/vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
config.json
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "facebook/m2m100_418M",
|
| 3 |
+
"activation_dropout": 0.0,
|
| 4 |
+
"activation_function": "relu",
|
| 5 |
+
"architectures": [
|
| 6 |
+
"M2M100ForConditionalGeneration"
|
| 7 |
+
],
|
| 8 |
+
"attention_dropout": 0.1,
|
| 9 |
+
"bos_token_id": 0,
|
| 10 |
+
"d_model": 1024,
|
| 11 |
+
"decoder_attention_heads": 16,
|
| 12 |
+
"decoder_ffn_dim": 4096,
|
| 13 |
+
"decoder_layerdrop": 0.05,
|
| 14 |
+
"decoder_layers": 12,
|
| 15 |
+
"decoder_start_token_id": 2,
|
| 16 |
+
"dropout": 0.1,
|
| 17 |
+
"early_stopping": true,
|
| 18 |
+
"encoder_attention_heads": 16,
|
| 19 |
+
"encoder_ffn_dim": 4096,
|
| 20 |
+
"encoder_layerdrop": 0.05,
|
| 21 |
+
"encoder_layers": 12,
|
| 22 |
+
"eos_token_id": 2,
|
| 23 |
+
"gradient_checkpointing": false,
|
| 24 |
+
"init_std": 0.02,
|
| 25 |
+
"is_encoder_decoder": true,
|
| 26 |
+
"max_length": 200,
|
| 27 |
+
"max_position_embeddings": 1024,
|
| 28 |
+
"model_type": "m2m_100",
|
| 29 |
+
"num_beams": 5,
|
| 30 |
+
"num_hidden_layers": 12,
|
| 31 |
+
"pad_token_id": 1,
|
| 32 |
+
"scale_embedding": true,
|
| 33 |
+
"torch_dtype": "float32",
|
| 34 |
+
"transformers_version": "4.28.0",
|
| 35 |
+
"use_cache": true,
|
| 36 |
+
"vocab_size": 128112
|
| 37 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 0,
|
| 3 |
+
"decoder_start_token_id": 2,
|
| 4 |
+
"early_stopping": true,
|
| 5 |
+
"eos_token_id": 2,
|
| 6 |
+
"max_length": 200,
|
| 7 |
+
"num_beams": 5,
|
| 8 |
+
"pad_token_id": 1,
|
| 9 |
+
"transformers_version": "4.28.0"
|
| 10 |
+
}
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c353a0da2bdb44caf3bc774a1f09200622a89d1206338b7f9ac554bd3cd9171d
|
| 3 |
+
size 1944201353
|
sentencepiece.bpe.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d8f7c76ed2a5e0822be39f0a4f95a55eb19c78f4593ce609e2edbc2aea4d380a
|
| 3 |
+
size 2423393
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,109 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"__af__",
|
| 4 |
+
"__am__",
|
| 5 |
+
"__ar__",
|
| 6 |
+
"__ast__",
|
| 7 |
+
"__az__",
|
| 8 |
+
"__ba__",
|
| 9 |
+
"__be__",
|
| 10 |
+
"__bg__",
|
| 11 |
+
"__bn__",
|
| 12 |
+
"__br__",
|
| 13 |
+
"__bs__",
|
| 14 |
+
"__ca__",
|
| 15 |
+
"__ceb__",
|
| 16 |
+
"__cs__",
|
| 17 |
+
"__cy__",
|
| 18 |
+
"__da__",
|
| 19 |
+
"__de__",
|
| 20 |
+
"__el__",
|
| 21 |
+
"__en__",
|
| 22 |
+
"__es__",
|
| 23 |
+
"__et__",
|
| 24 |
+
"__fa__",
|
| 25 |
+
"__ff__",
|
| 26 |
+
"__fi__",
|
| 27 |
+
"__fr__",
|
| 28 |
+
"__fy__",
|
| 29 |
+
"__ga__",
|
| 30 |
+
"__gd__",
|
| 31 |
+
"__gl__",
|
| 32 |
+
"__gu__",
|
| 33 |
+
"__ha__",
|
| 34 |
+
"__he__",
|
| 35 |
+
"__hi__",
|
| 36 |
+
"__hr__",
|
| 37 |
+
"__ht__",
|
| 38 |
+
"__hu__",
|
| 39 |
+
"__hy__",
|
| 40 |
+
"__id__",
|
| 41 |
+
"__ig__",
|
| 42 |
+
"__ilo__",
|
| 43 |
+
"__is__",
|
| 44 |
+
"__it__",
|
| 45 |
+
"__ja__",
|
| 46 |
+
"__jv__",
|
| 47 |
+
"__ka__",
|
| 48 |
+
"__kk__",
|
| 49 |
+
"__km__",
|
| 50 |
+
"__kn__",
|
| 51 |
+
"__ko__",
|
| 52 |
+
"__lb__",
|
| 53 |
+
"__lg__",
|
| 54 |
+
"__ln__",
|
| 55 |
+
"__lo__",
|
| 56 |
+
"__lt__",
|
| 57 |
+
"__lv__",
|
| 58 |
+
"__mg__",
|
| 59 |
+
"__mk__",
|
| 60 |
+
"__ml__",
|
| 61 |
+
"__mn__",
|
| 62 |
+
"__mr__",
|
| 63 |
+
"__ms__",
|
| 64 |
+
"__my__",
|
| 65 |
+
"__ne__",
|
| 66 |
+
"__nl__",
|
| 67 |
+
"__no__",
|
| 68 |
+
"__ns__",
|
| 69 |
+
"__oc__",
|
| 70 |
+
"__or__",
|
| 71 |
+
"__pa__",
|
| 72 |
+
"__pl__",
|
| 73 |
+
"__ps__",
|
| 74 |
+
"__pt__",
|
| 75 |
+
"__ro__",
|
| 76 |
+
"__ru__",
|
| 77 |
+
"__sd__",
|
| 78 |
+
"__si__",
|
| 79 |
+
"__sk__",
|
| 80 |
+
"__sl__",
|
| 81 |
+
"__so__",
|
| 82 |
+
"__sq__",
|
| 83 |
+
"__sr__",
|
| 84 |
+
"__ss__",
|
| 85 |
+
"__su__",
|
| 86 |
+
"__sv__",
|
| 87 |
+
"__sw__",
|
| 88 |
+
"__ta__",
|
| 89 |
+
"__th__",
|
| 90 |
+
"__tl__",
|
| 91 |
+
"__tn__",
|
| 92 |
+
"__tr__",
|
| 93 |
+
"__uk__",
|
| 94 |
+
"__ur__",
|
| 95 |
+
"__uz__",
|
| 96 |
+
"__vi__",
|
| 97 |
+
"__wo__",
|
| 98 |
+
"__xh__",
|
| 99 |
+
"__yi__",
|
| 100 |
+
"__yo__",
|
| 101 |
+
"__zh__",
|
| 102 |
+
"__zu__"
|
| 103 |
+
],
|
| 104 |
+
"bos_token": "<s>",
|
| 105 |
+
"eos_token": "</s>",
|
| 106 |
+
"pad_token": "<pad>",
|
| 107 |
+
"sep_token": "</s>",
|
| 108 |
+
"unk_token": "<unk>"
|
| 109 |
+
}
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,118 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"__af__",
|
| 4 |
+
"__am__",
|
| 5 |
+
"__ar__",
|
| 6 |
+
"__ast__",
|
| 7 |
+
"__az__",
|
| 8 |
+
"__ba__",
|
| 9 |
+
"__be__",
|
| 10 |
+
"__bg__",
|
| 11 |
+
"__bn__",
|
| 12 |
+
"__br__",
|
| 13 |
+
"__bs__",
|
| 14 |
+
"__ca__",
|
| 15 |
+
"__ceb__",
|
| 16 |
+
"__cs__",
|
| 17 |
+
"__cy__",
|
| 18 |
+
"__da__",
|
| 19 |
+
"__de__",
|
| 20 |
+
"__el__",
|
| 21 |
+
"__en__",
|
| 22 |
+
"__es__",
|
| 23 |
+
"__et__",
|
| 24 |
+
"__fa__",
|
| 25 |
+
"__ff__",
|
| 26 |
+
"__fi__",
|
| 27 |
+
"__fr__",
|
| 28 |
+
"__fy__",
|
| 29 |
+
"__ga__",
|
| 30 |
+
"__gd__",
|
| 31 |
+
"__gl__",
|
| 32 |
+
"__gu__",
|
| 33 |
+
"__ha__",
|
| 34 |
+
"__he__",
|
| 35 |
+
"__hi__",
|
| 36 |
+
"__hr__",
|
| 37 |
+
"__ht__",
|
| 38 |
+
"__hu__",
|
| 39 |
+
"__hy__",
|
| 40 |
+
"__id__",
|
| 41 |
+
"__ig__",
|
| 42 |
+
"__ilo__",
|
| 43 |
+
"__is__",
|
| 44 |
+
"__it__",
|
| 45 |
+
"__ja__",
|
| 46 |
+
"__jv__",
|
| 47 |
+
"__ka__",
|
| 48 |
+
"__kk__",
|
| 49 |
+
"__km__",
|
| 50 |
+
"__kn__",
|
| 51 |
+
"__ko__",
|
| 52 |
+
"__lb__",
|
| 53 |
+
"__lg__",
|
| 54 |
+
"__ln__",
|
| 55 |
+
"__lo__",
|
| 56 |
+
"__lt__",
|
| 57 |
+
"__lv__",
|
| 58 |
+
"__mg__",
|
| 59 |
+
"__mk__",
|
| 60 |
+
"__ml__",
|
| 61 |
+
"__mn__",
|
| 62 |
+
"__mr__",
|
| 63 |
+
"__ms__",
|
| 64 |
+
"__my__",
|
| 65 |
+
"__ne__",
|
| 66 |
+
"__nl__",
|
| 67 |
+
"__no__",
|
| 68 |
+
"__ns__",
|
| 69 |
+
"__oc__",
|
| 70 |
+
"__or__",
|
| 71 |
+
"__pa__",
|
| 72 |
+
"__pl__",
|
| 73 |
+
"__ps__",
|
| 74 |
+
"__pt__",
|
| 75 |
+
"__ro__",
|
| 76 |
+
"__ru__",
|
| 77 |
+
"__sd__",
|
| 78 |
+
"__si__",
|
| 79 |
+
"__sk__",
|
| 80 |
+
"__sl__",
|
| 81 |
+
"__so__",
|
| 82 |
+
"__sq__",
|
| 83 |
+
"__sr__",
|
| 84 |
+
"__ss__",
|
| 85 |
+
"__su__",
|
| 86 |
+
"__sv__",
|
| 87 |
+
"__sw__",
|
| 88 |
+
"__ta__",
|
| 89 |
+
"__th__",
|
| 90 |
+
"__tl__",
|
| 91 |
+
"__tn__",
|
| 92 |
+
"__tr__",
|
| 93 |
+
"__uk__",
|
| 94 |
+
"__ur__",
|
| 95 |
+
"__uz__",
|
| 96 |
+
"__vi__",
|
| 97 |
+
"__wo__",
|
| 98 |
+
"__xh__",
|
| 99 |
+
"__yi__",
|
| 100 |
+
"__yo__",
|
| 101 |
+
"__zh__",
|
| 102 |
+
"__zu__"
|
| 103 |
+
],
|
| 104 |
+
"bos_token": "<s>",
|
| 105 |
+
"clean_up_tokenization_spaces": true,
|
| 106 |
+
"eos_token": "</s>",
|
| 107 |
+
"language_codes": "m2m100",
|
| 108 |
+
"model_max_length": 1024,
|
| 109 |
+
"num_madeup_words": 8,
|
| 110 |
+
"pad_token": "<pad>",
|
| 111 |
+
"sep_token": "</s>",
|
| 112 |
+
"sp_model_kwargs": {},
|
| 113 |
+
"src_lang": "en",
|
| 114 |
+
"tgt_lang": "fr",
|
| 115 |
+
"tokenizer_class": "M2M100Tokenizer",
|
| 116 |
+
"tokenizer_file": null,
|
| 117 |
+
"unk_token": "<unk>"
|
| 118 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:98438687991f47832564d8ab84087cb1c5a15d8246397eca599954545732c6e4
|
| 3 |
+
size 3771
|
vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|