Ali Safaya commited on
Commit
fc7b92b
1 Parent(s): 6448b7b

initial commit

Browse files
README.md CHANGED
@@ -1,3 +1,91 @@
1
- ---
2
- license: cc
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - generated_from_trainer
4
+ datasets:
5
+ - mlsum
6
+ metrics:
7
+ - rouge
8
+ model-index:
9
+ - name: eval-mbart-large
10
+ results:
11
+ - task:
12
+ name: Summarization
13
+ type: summarization
14
+ dataset:
15
+ name: mlsum tu
16
+ type: mlsum
17
+ args: tu
18
+ metrics:
19
+ - name: Rouge1
20
+ type: rouge
21
+ value: 46.7011
22
+ ---
23
+
24
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
25
+ should probably proofread and complete it, then remove this comment. -->
26
+
27
+ # eval-mbart-large
28
+
29
+ This model is a fine-tuned version of [facebook/mbart-large-50](https://huggingface.co/facebook/mbart-large-50) on the mlsum tu dataset.
30
+ It achieves the following results on the evaluation set:
31
+ - Loss: 2.8386
32
+ - Rouge1: 46.7011
33
+ - Rouge2: 34.0087
34
+ - Rougel: 41.5475
35
+ - Rougelsum: 43.2108
36
+ - Gen Len: 43.2426
37
+
38
+ ## Model description
39
+
40
+ More information needed
41
+
42
+ ## Intended uses & limitations
43
+
44
+ More information needed
45
+
46
+ ## Training and evaluation data
47
+
48
+ More information needed
49
+
50
+ ## Training procedure
51
+
52
+ ### Training hyperparameters
53
+
54
+ The following hyperparameters were used during training:
55
+ - learning_rate: 5e-05
56
+ - train_batch_size: 2
57
+ - eval_batch_size: 4
58
+ - seed: 42
59
+ - distributed_type: multi-GPU
60
+ - num_devices: 8
61
+ - gradient_accumulation_steps: 4
62
+ - total_train_batch_size: 64
63
+ - total_eval_batch_size: 32
64
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
65
+ - lr_scheduler_type: linear
66
+ - num_epochs: 10.0
67
+ - mixed_precision_training: Native AMP
68
+ - label_smoothing_factor: 0.1
69
+
70
+ ### Training results
71
+
72
+ | Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
73
+ |:-------------:|:-----:|:-----:|:---------------:|:-------:|:-------:|:-------:|:---------:|:-------:|
74
+ | 2.9098 | 1.0 | 3895 | 2.8820 | 45.2085 | 33.0253 | 40.3511 | 41.8378 | 40.1802 |
75
+ | 2.7496 | 2.0 | 7790 | 2.8620 | 45.3455 | 33.0049 | 40.4574 | 41.9738 | 40.845 |
76
+ | 2.6215 | 3.0 | 11685 | 2.8386 | 46.6642 | 34.0133 | 41.5102 | 43.1852 | 43.3505 |
77
+ | 2.5031 | 4.0 | 15580 | 2.8620 | 46.5081 | 33.9028 | 41.5001 | 43.0841 | 42.4534 |
78
+ | 2.3967 | 5.0 | 19475 | 2.8935 | 46.1029 | 33.4495 | 41.0557 | 42.7096 | 41.9169 |
79
+ | 2.3161 | 6.0 | 23370 | 2.9255 | 46.0193 | 33.2904 | 40.9323 | 42.575 | 42.5379 |
80
+ | 2.2348 | 7.0 | 27265 | 2.9692 | 46.4242 | 33.718 | 41.4037 | 43.0504 | 41.7957 |
81
+ | 2.1716 | 8.0 | 31160 | 3.0044 | 46.1669 | 33.3276 | 40.9307 | 42.7015 | 42.8942 |
82
+ | 2.1179 | 9.0 | 35055 | 3.0372 | 46.0666 | 33.2483 | 40.9372 | 42.6837 | 42.7636 |
83
+ | 2.0753 | 10.0 | 38950 | 3.0627 | 46.1444 | 33.2551 | 40.9514 | 42.7096 | 42.9266 |
84
+
85
+
86
+ ### Framework versions
87
+
88
+ - Transformers 4.11.3
89
+ - Pytorch 1.8.2+cu111
90
+ - Datasets 1.14.0
91
+ - Tokenizers 0.10.3
config.json ADDED
@@ -0,0 +1,59 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "facebook/mbart-large-50",
3
+ "_num_labels": 3,
4
+ "activation_dropout": 0.0,
5
+ "activation_function": "gelu",
6
+ "add_bias_logits": false,
7
+ "add_final_layer_norm": true,
8
+ "architectures": [
9
+ "MBartForConditionalGeneration"
10
+ ],
11
+ "attention_dropout": 0.0,
12
+ "bos_token_id": 0,
13
+ "classif_dropout": 0.0,
14
+ "classifier_dropout": 0.0,
15
+ "d_model": 1024,
16
+ "decoder_attention_heads": 16,
17
+ "decoder_ffn_dim": 4096,
18
+ "decoder_layerdrop": 0.0,
19
+ "decoder_layers": 12,
20
+ "decoder_start_token_id": 2,
21
+ "dropout": 0.1,
22
+ "early_stopping": true,
23
+ "encoder_attention_heads": 16,
24
+ "encoder_ffn_dim": 4096,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 12,
27
+ "eos_token_id": 2,
28
+ "forced_bos_token_id": 250023,
29
+ "forced_eos_token_id": 2,
30
+ "gradient_checkpointing": false,
31
+ "id2label": {
32
+ "0": "LABEL_0",
33
+ "1": "LABEL_1",
34
+ "2": "LABEL_2"
35
+ },
36
+ "init_std": 0.02,
37
+ "is_encoder_decoder": true,
38
+ "label2id": {
39
+ "LABEL_0": 0,
40
+ "LABEL_1": 1,
41
+ "LABEL_2": 2
42
+ },
43
+ "max_length": 200,
44
+ "max_position_embeddings": 1024,
45
+ "model_type": "mbart",
46
+ "normalize_before": true,
47
+ "normalize_embedding": true,
48
+ "num_beams": 5,
49
+ "num_hidden_layers": 12,
50
+ "output_past": true,
51
+ "pad_token_id": 1,
52
+ "scale_embedding": true,
53
+ "static_position_embeddings": false,
54
+ "tokenizer_class": "MBart50Tokenizer",
55
+ "torch_dtype": "float32",
56
+ "transformers_version": "4.11.3",
57
+ "use_cache": true,
58
+ "vocab_size": 250054
59
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ae44ca016a350017780d65061173081b121bae883bec55fd702a86edb70d18d6
3
+ size 2444722195
sentencepiece.bpe.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cfc8146abe2a0488e9e2a0c56de7952f7c11ab059eca145a0a727afce0db2865
3
+ size 5069051
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "sep_token": "</s>", "pad_token": "<pad>", "cls_token": "<s>", "mask_token": "<mask>", "additional_special_tokens": ["ar_AR", "cs_CZ", "de_DE", "en_XX", "es_XX", "et_EE", "fi_FI", "fr_XX", "gu_IN", "hi_IN", "it_IT", "ja_XX", "kk_KZ", "ko_KR", "lt_LT", "lv_LV", "my_MM", "ne_NP", "nl_XX", "ro_RO", "ru_RU", "si_LK", "tr_TR", "vi_VN", "zh_CN", "af_ZA", "az_AZ", "bn_IN", "fa_IR", "he_IL", "hr_HR", "id_ID", "ka_GE", "km_KH", "mk_MK", "ml_IN", "mn_MN", "mr_IN", "pl_PL", "ps_AF", "pt_XX", "sv_SE", "sw_KE", "ta_IN", "te_IN", "th_TH", "tl_XX", "uk_UA", "ur_PK", "xh_ZA", "gl_ES", "sl_SI"]}
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"src_lang": null, "tgt_lang": null, "eos_token": "</s>", "sep_token": "</s>", "cls_token": "<s>", "unk_token": "<unk>", "pad_token": "<pad>", "mask_token": {"content": "<mask>", "single_word": false, "lstrip": true, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "bos_token": "<s>", "model_max_length": 1024, "name_or_path": "facebook/mbart-large-50", "special_tokens_map_file": "/home/suraj/projects/mbart-50/hf_models/mbart-50-large/special_tokens_map.json", "additional_special_tokens": ["ar_AR", "cs_CZ", "de_DE", "en_XX", "es_XX", "et_EE", "fi_FI", "fr_XX", "gu_IN", "hi_IN", "it_IT", "ja_XX", "kk_KZ", "ko_KR", "lt_LT", "lv_LV", "my_MM", "ne_NP", "nl_XX", "ro_RO", "ru_RU", "si_LK", "tr_TR", "vi_VN", "zh_CN", "af_ZA", "az_AZ", "bn_IN", "fa_IR", "he_IL", "hr_HR", "id_ID", "ka_GE", "km_KH", "mk_MK", "ml_IN", "mn_MN", "mr_IN", "pl_PL", "ps_AF", "pt_XX", "sv_SE", "sw_KE", "ta_IN", "te_IN", "th_TH", "tl_XX", "uk_UA", "ur_PK", "xh_ZA", "gl_ES", "sl_SI"], "sp_model_kwargs": {}, "tokenizer_class": "MBart50Tokenizer"}