lekhnathrijal commited on
Commit
9fbca95
·
verified ·
1 Parent(s): 33bc7d7

ai-research-lab/bert-question-classifier

Browse files
Files changed (4) hide show
  1. README.md +29 -37
  2. config.json +18 -0
  3. model.safetensors +1 -1
  4. tokenizer.json +10 -1
README.md CHANGED
@@ -21,11 +21,11 @@ should probably proofread and complete it, then remove this comment. -->
21
 
22
  This model is a fine-tuned version of [google-bert/bert-base-uncased](https://huggingface.co/google-bert/bert-base-uncased) on the None dataset.
23
  It achieves the following results on the evaluation set:
24
- - Loss: 1.6111
25
- - Accuracy: 0.9727
26
- - Recall: 0.8589
27
- - Precision: 0.8434
28
- - F1: 0.8511
29
 
30
  ## Model description
31
 
@@ -51,43 +51,35 @@ The following hyperparameters were used during training:
51
  - optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
52
  - lr_scheduler_type: linear
53
  - lr_scheduler_warmup_ratio: 0.1
54
- - num_epochs: 8
55
 
56
  ### Training results
57
 
58
  | Training Loss | Epoch | Step | Validation Loss | Accuracy | Recall | Precision | F1 |
59
  |:-------------:|:------:|:----:|:---------------:|:--------:|:------:|:---------:|:------:|
60
- | No log | 0.0959 | 100 | 5.6110 | 0.8592 | 0.0493 | 0.0762 | 0.0599 |
61
- | No log | 0.1918 | 200 | 5.1773 | 0.9031 | 0.4442 | 0.4654 | 0.4545 |
62
- | No log | 0.2876 | 300 | 4.9323 | 0.9166 | 0.5264 | 0.5423 | 0.5342 |
63
- | No log | 0.3835 | 400 | 4.6288 | 0.9227 | 0.5602 | 0.5774 | 0.5686 |
64
- | 5.1489 | 0.4794 | 500 | 4.3343 | 0.9287 | 0.5934 | 0.6114 | 0.6022 |
65
- | 5.1489 | 0.5753 | 600 | 4.0298 | 0.9326 | 0.6281 | 0.6295 | 0.6288 |
66
- | 5.1489 | 0.6711 | 700 | 3.8647 | 0.9338 | 0.6414 | 0.6346 | 0.6380 |
67
- | 5.1489 | 0.7670 | 800 | 3.5865 | 0.9425 | 0.6991 | 0.6785 | 0.6887 |
68
- | 5.1489 | 0.8629 | 900 | 3.4043 | 0.9461 | 0.7156 | 0.6985 | 0.7069 |
69
- | 3.7588 | 0.9588 | 1000 | 3.2239 | 0.9500 | 0.7249 | 0.7253 | 0.7251 |
70
- | 3.7588 | 1.0547 | 1100 | 3.0973 | 0.9503 | 0.7305 | 0.7253 | 0.7279 |
71
- | 3.7588 | 1.1505 | 1200 | 2.9045 | 0.9516 | 0.7441 | 0.7287 | 0.7363 |
72
- | 3.7588 | 1.2464 | 1300 | 2.7053 | 0.9561 | 0.7683 | 0.7536 | 0.7609 |
73
- | 3.7588 | 1.3423 | 1400 | 2.5678 | 0.9585 | 0.7819 | 0.7663 | 0.7740 |
74
- | 2.8585 | 1.4382 | 1500 | 2.4580 | 0.9594 | 0.7841 | 0.7729 | 0.7784 |
75
- | 2.8585 | 1.5340 | 1600 | 2.3973 | 0.9589 | 0.7872 | 0.7670 | 0.7770 |
76
- | 2.8585 | 1.6299 | 1700 | 2.2791 | 0.9622 | 0.8061 | 0.7838 | 0.7948 |
77
- | 2.8585 | 1.7258 | 1800 | 2.2024 | 0.9645 | 0.8176 | 0.7973 | 0.8074 |
78
- | 2.8585 | 1.8217 | 1900 | 2.1277 | 0.9640 | 0.8161 | 0.7939 | 0.8048 |
79
- | 2.2551 | 1.9175 | 2000 | 2.0425 | 0.9660 | 0.8229 | 0.8071 | 0.8149 |
80
- | 2.2551 | 2.0134 | 2100 | 1.9677 | 0.9670 | 0.8306 | 0.8108 | 0.8206 |
81
- | 2.2551 | 2.1093 | 2200 | 1.9116 | 0.9672 | 0.8303 | 0.8129 | 0.8215 |
82
- | 2.2551 | 2.2052 | 2300 | 1.8522 | 0.9680 | 0.8337 | 0.8178 | 0.8257 |
83
- | 2.2551 | 2.3011 | 2400 | 1.8227 | 0.9684 | 0.8378 | 0.8187 | 0.8281 |
84
- | 1.8456 | 2.3969 | 2500 | 1.7578 | 0.9699 | 0.8415 | 0.8297 | 0.8355 |
85
- | 1.8456 | 2.4928 | 2600 | 1.7359 | 0.9703 | 0.8505 | 0.8276 | 0.8389 |
86
- | 1.8456 | 2.5887 | 2700 | 1.6956 | 0.9704 | 0.8502 | 0.8283 | 0.8391 |
87
- | 1.8456 | 2.6846 | 2800 | 1.6397 | 0.9711 | 0.8539 | 0.8327 | 0.8432 |
88
- | 1.8456 | 2.7804 | 2900 | 1.6111 | 0.9727 | 0.8589 | 0.8434 | 0.8511 |
89
- | 1.5842 | 2.8763 | 3000 | 1.5852 | 0.9726 | 0.8607 | 0.8412 | 0.8508 |
90
- | 1.5842 | 2.9722 | 3100 | 1.5659 | 0.9723 | 0.8536 | 0.8431 | 0.8483 |
91
 
92
 
93
  ### Framework versions
 
21
 
22
  This model is a fine-tuned version of [google-bert/bert-base-uncased](https://huggingface.co/google-bert/bert-base-uncased) on the None dataset.
23
  It achieves the following results on the evaluation set:
24
+ - Loss: 1.9349
25
+ - Accuracy: 0.9687
26
+ - Recall: 0.8393
27
+ - Precision: 0.8205
28
+ - F1: 0.8298
29
 
30
  ## Model description
31
 
 
51
  - optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
52
  - lr_scheduler_type: linear
53
  - lr_scheduler_warmup_ratio: 0.1
54
+ - num_epochs: 10
55
 
56
  ### Training results
57
 
58
  | Training Loss | Epoch | Step | Validation Loss | Accuracy | Recall | Precision | F1 |
59
  |:-------------:|:------:|:----:|:---------------:|:--------:|:------:|:---------:|:------:|
60
+ | No log | 0.0959 | 100 | 5.3220 | 0.8786 | 0.3772 | 0.3462 | 0.3610 |
61
+ | No log | 0.1918 | 200 | 5.1118 | 0.9118 | 0.5105 | 0.5152 | 0.5129 |
62
+ | No log | 0.2876 | 300 | 4.7630 | 0.9237 | 0.5934 | 0.5786 | 0.5859 |
63
+ | No log | 0.3835 | 400 | 4.5162 | 0.9247 | 0.6079 | 0.5821 | 0.5948 |
64
+ | 4.9675 | 0.4794 | 500 | 4.2140 | 0.9288 | 0.6300 | 0.6041 | 0.6168 |
65
+ | 4.9675 | 0.5753 | 600 | 3.9853 | 0.9357 | 0.6694 | 0.6402 | 0.6544 |
66
+ | 4.9675 | 0.6711 | 700 | 3.7985 | 0.9364 | 0.6600 | 0.6476 | 0.6538 |
67
+ | 4.9675 | 0.7670 | 800 | 3.5336 | 0.9426 | 0.7044 | 0.6775 | 0.6907 |
68
+ | 4.9675 | 0.8629 | 900 | 3.3734 | 0.9438 | 0.7094 | 0.6841 | 0.6965 |
69
+ | 3.7188 | 0.9588 | 1000 | 3.1605 | 0.9482 | 0.7326 | 0.7078 | 0.7200 |
70
+ | 3.7188 | 1.0547 | 1100 | 3.0154 | 0.9495 | 0.7317 | 0.7181 | 0.7248 |
71
+ | 3.7188 | 1.1505 | 1200 | 2.8699 | 0.9518 | 0.7478 | 0.7293 | 0.7384 |
72
+ | 3.7188 | 1.2464 | 1300 | 2.7020 | 0.9540 | 0.7605 | 0.7406 | 0.7504 |
73
+ | 3.7188 | 1.3423 | 1400 | 2.5487 | 0.9587 | 0.7863 | 0.7654 | 0.7757 |
74
+ | 2.8304 | 1.4382 | 1500 | 2.4370 | 0.9609 | 0.8027 | 0.7749 | 0.7885 |
75
+ | 2.8304 | 1.5340 | 1600 | 2.3423 | 0.9625 | 0.8043 | 0.7874 | 0.7958 |
76
+ | 2.8304 | 1.6299 | 1700 | 2.2367 | 0.9637 | 0.8117 | 0.7940 | 0.8028 |
77
+ | 2.8304 | 1.7258 | 1800 | 2.1771 | 0.9652 | 0.8154 | 0.8045 | 0.8099 |
78
+ | 2.8304 | 1.8217 | 1900 | 2.0955 | 0.9655 | 0.8248 | 0.8019 | 0.8131 |
79
+ | 2.2139 | 1.9175 | 2000 | 2.0031 | 0.9676 | 0.8316 | 0.8159 | 0.8237 |
80
+ | 2.2139 | 2.0134 | 2100 | 1.9349 | 0.9687 | 0.8393 | 0.8205 | 0.8298 |
81
+ | 2.2139 | 2.1093 | 2200 | 1.8722 | 0.9683 | 0.8344 | 0.8201 | 0.8272 |
82
+ | 2.2139 | 2.2052 | 2300 | 1.8385 | 0.9682 | 0.8396 | 0.8161 | 0.8277 |
 
 
 
 
 
 
 
 
83
 
84
 
85
  ### Framework versions
config.json CHANGED
@@ -5,6 +5,24 @@
5
  ],
6
  "attention_probs_dropout_prob": 0.1,
7
  "classifier_dropout": null,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8
  "gradient_checkpointing": false,
9
  "hidden_act": "gelu",
10
  "hidden_dropout_prob": 0.1,
 
5
  ],
6
  "attention_probs_dropout_prob": 0.1,
7
  "classifier_dropout": null,
8
+ "custom_pipelines": {
9
+ "question-classifier": {
10
+ "default": {
11
+ "model": {
12
+ "pt": [
13
+ "ai-research-lab/bert-question-classifier",
14
+ "main"
15
+ ]
16
+ }
17
+ },
18
+ "impl": "classifier_pipeline.MultiTaskClassifierPipeline",
19
+ "pt": [
20
+ "AutoModelForSequenceClassification"
21
+ ],
22
+ "tf": [],
23
+ "type": "text"
24
+ }
25
+ },
26
  "gradient_checkpointing": false,
27
  "hidden_act": "gelu",
28
  "hidden_dropout_prob": 0.1,
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e4acafb8631fdee9d0fe509662a1f0c80d7361277e182a3519582fca8023ddf3
3
  size 438057080
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a57f19aad1305b29947aa312b1f7dbdd6b9a9a5c6a139891711ae14d2ec3139b
3
  size 438057080
tokenizer.json CHANGED
@@ -6,7 +6,16 @@
6
  "strategy": "LongestFirst",
7
  "stride": 0
8
  },
9
- "padding": null,
 
 
 
 
 
 
 
 
 
10
  "added_tokens": [
11
  {
12
  "id": 0,
 
6
  "strategy": "LongestFirst",
7
  "stride": 0
8
  },
9
+ "padding": {
10
+ "strategy": {
11
+ "Fixed": 512
12
+ },
13
+ "direction": "Right",
14
+ "pad_to_multiple_of": null,
15
+ "pad_id": 0,
16
+ "pad_type_id": 0,
17
+ "pad_token": "[PAD]"
18
+ },
19
  "added_tokens": [
20
  {
21
  "id": 0,