Hugging Face's logo

kaixkhazaki
/
turkish-question-answering

Question Answering
Transformers
Safetensors
Turkish
bert
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Community
kaixkhazaki commited on
Commit
a8ec187
·
verified ·
1 Parent(s): 6784d06

Pushing of the new best model checkpoint

Browse files
Files changed (4) hide show
  1. README.md +75 -71
  2. model.safetensors +1 -1
  3. tokenizer.json +2 -16
  4. training_args.bin +1 -1
README.md CHANGED
@@ -23,10 +23,7 @@ model-index:
23
  metrics:
24
  - name: F1
25
  type: f1
26
- value: 62.81121459854913
27
- language:
28
- - tr
29
- pipeline_tag: question-answering
30
  ---
31
 
32
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -36,18 +33,9 @@ should probably proofread and complete it, then remove this comment. -->
36
 
37
  This model is a fine-tuned version of [dbmdz/bert-base-turkish-cased](https://huggingface.co/dbmdz/bert-base-turkish-cased) on the squad_tr dataset.
38
  It achieves the following results on the evaluation set:
39
- - Loss: 1.34
40
- - Exact Match: 51.52
41
- - F1: 62.81
42
-
43
- {'eval_loss': 1.3433560132980347,
44
- 'eval_exact_match': 51.528998242530754,
45
- 'eval_f1': 62.81121459854913,
46
- 'eval_runtime': 26.378,
47
- 'eval_samples_per_second': 402.153,
48
- 'eval_steps_per_second': 6.293,
49
- 'epoch': 1.74321153201475}
50
-
51
 
52
  ## Model description
53
 
@@ -66,12 +54,12 @@ More information needed
66
  ### Training hyperparameters
67
 
68
  The following hyperparameters were used during training:
69
- - learning_rate: 5e-05
70
  - train_batch_size: 16
71
  - eval_batch_size: 64
72
  - seed: 42
73
  - optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
74
- - lr_scheduler_type: cosine
75
  - lr_scheduler_warmup_steps: 500
76
  - num_epochs: 3
77
 
@@ -79,58 +67,74 @@ The following hyperparameters were used during training:
79
 
80
  | Training Loss | Epoch | Step | Validation Loss | Exact Match | F1 |
81
  |:-------------:|:------:|:-----:|:---------------:|:-----------:|:-------:|
82
- | 2.748 | 0.0335 | 200 | 2.4802 | 2.5708 | 3.3209 |
83
- | 2.2327 | 0.0670 | 400 | 2.0295 | 14.1005 | 17.3881 |
84
- | 1.9586 | 0.1006 | 600 | 1.9123 | 18.0192 | 22.5210 |
85
- | 1.864 | 0.1341 | 800 | 1.7851 | 26.4133 | 31.0023 |
86
- | 1.8589 | 0.1676 | 1000 | 1.7175 | 30.0436 | 36.0232 |
87
- | 1.6896 | 0.2011 | 1200 | 1.6589 | 27.9475 | 32.8651 |
88
- | 1.7921 | 0.2347 | 1400 | 1.6032 | 34.3599 | 41.7296 |
89
- | 1.6259 | 0.2682 | 1600 | 1.5761 | 33.7665 | 41.5212 |
90
- | 1.6319 | 0.3017 | 1800 | 1.5520 | 34.6606 | 41.7878 |
91
- | 1.5926 | 0.3352 | 2000 | 1.5287 | 35.9920 | 43.3277 |
92
- | 1.5903 | 0.3688 | 2200 | 1.4975 | 37.9788 | 45.6546 |
93
- | 1.5358 | 0.4023 | 2400 | 1.4915 | 33.6006 | 40.5064 |
94
- | 1.4999 | 0.4358 | 2600 | 1.5199 | 30.3836 | 36.6080 |
95
- | 1.4952 | 0.4693 | 2800 | 1.4471 | 41.2024 | 50.0649 |
96
- | 1.4307 | 0.5028 | 3000 | 1.4574 | 41.7014 | 51.9319 |
97
- | 1.5735 | 0.5364 | 3200 | 1.4210 | 41.3499 | 50.8481 |
98
- | 1.4442 | 0.5699 | 3400 | 1.3738 | 44.7996 | 54.8485 |
99
- | 1.3585 | 0.6034 | 3600 | 1.3800 | 42.6988 | 52.4796 |
100
- | 1.3796 | 0.6369 | 3800 | 1.3631 | 43.3681 | 52.9250 |
101
- | 1.4985 | 0.6705 | 4000 | 1.3416 | 41.1262 | 50.4030 |
102
- | 1.3423 | 0.7040 | 4200 | 1.4062 | 34.0532 | 41.4662 |
103
- | 1.3888 | 0.7375 | 4400 | 1.4006 | 46.5225 | 57.5515 |
104
- | 1.4313 | 0.7710 | 4600 | 1.3052 | 45.4704 | 55.5428 |
105
- | 1.3249 | 0.8046 | 4800 | 1.3356 | 43.1737 | 52.1946 |
106
- | 1.3725 | 0.8381 | 5000 | 1.3371 | 48.0294 | 60.2390 |
107
- | 1.4246 | 0.8716 | 5200 | 1.2925 | 42.0401 | 51.2726 |
108
- | 1.4384 | 0.9051 | 5400 | 1.2820 | 40.7329 | 49.4549 |
109
- | 1.326 | 0.9387 | 5600 | 1.3022 | 46.9399 | 57.0599 |
110
- | 1.4134 | 0.9722 | 5800 | 1.2620 | 45.2739 | 55.2044 |
111
- | 1.2021 | 1.0057 | 6000 | 1.3326 | 47.0021 | 57.5404 |
112
- | 0.9818 | 1.0392 | 6200 | 1.3769 | 42.5416 | 51.1673 |
113
- | 0.9539 | 1.0727 | 6400 | 1.4191 | 45.7006 | 55.2988 |
114
- | 0.9542 | 1.1063 | 6600 | 1.4101 | 42.5418 | 51.3564 |
115
- | 1.0471 | 1.1398 | 6800 | 1.3400 | 48.3473 | 59.1064 |
116
- | 0.9518 | 1.1733 | 7000 | 1.3427 | 45.2902 | 55.9998 |
117
- | 1.041 | 1.2068 | 7200 | 1.3167 | 46.3732 | 56.7874 |
118
- | 1.0539 | 1.2404 | 7400 | 1.3434 | 51.5290 | 62.8112 |
119
- | 0.9751 | 1.2739 | 7600 | 1.3042 | 48.1280 | 59.1062 |
120
- | 1.0192 | 1.3074 | 7800 | 1.3235 | 40.6521 | 50.0644 |
121
- | 1.0309 | 1.3409 | 8000 | 1.3291 | 49.8233 | 61.3538 |
122
- | 1.0172 | 1.3745 | 8200 | 1.3165 | 47.7918 | 58.3036 |
123
- | 1.0202 | 1.4080 | 8400 | 1.2967 | 43.9059 | 54.1001 |
124
- | 0.9162 | 1.4415 | 8600 | 1.3100 | 43.1116 | 53.1443 |
125
- | 0.9833 | 1.4750 | 8800 | 1.2765 | 49.0851 | 59.7894 |
126
- | 1.0434 | 1.5085 | 9000 | 1.2699 | 48.7813 | 59.4381 |
127
- | 0.9666 | 1.5421 | 9200 | 1.2875 | 46.9806 | 57.2602 |
128
- | 0.9537 | 1.5756 | 9400 | 1.2711 | 46.2930 | 56.4741 |
129
- | 1.0466 | 1.6091 | 9600 | 1.2684 | 48.2181 | 58.9984 |
130
- | 1.0211 | 1.6426 | 9800 | 1.2364 | 48.2130 | 58.8486 |
131
- | 0.9688 | 1.6762 | 10000 | 1.3011 | 50.5397 | 61.6295 |
132
- | 0.9645 | 1.7097 | 10200 | 1.2577 | 50.1051 | 61.4147 |
133
- | 1.0133 | 1.7432 | 10400 | 1.2467 | 49.6075 | 60.7184 |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
134
 
135
 
136
  ### Framework versions
@@ -138,4 +142,4 @@ The following hyperparameters were used during training:
138
  - Transformers 4.48.0.dev0
139
  - Pytorch 2.4.1+cu121
140
  - Datasets 2.20.0
141
- - Tokenizers 0.21.0
 
23
  metrics:
24
  - name: F1
25
  type: f1
26
+ value: 56.95622375724343
 
 
 
27
  ---
28
 
29
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
33
 
34
  This model is a fine-tuned version of [dbmdz/bert-base-turkish-cased](https://huggingface.co/dbmdz/bert-base-turkish-cased) on the squad_tr dataset.
35
  It achieves the following results on the evaluation set:
36
+ - Loss: 1.3700
37
+ - Exact Match: 46.4242
38
+ - F1: 56.9562
 
 
 
 
 
 
 
 
 
39
 
40
  ## Model description
41
 
 
54
  ### Training hyperparameters
55
 
56
  The following hyperparameters were used during training:
57
+ - learning_rate: 3e-05
58
  - train_batch_size: 16
59
  - eval_batch_size: 64
60
  - seed: 42
61
  - optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
62
+ - lr_scheduler_type: linear
63
  - lr_scheduler_warmup_steps: 500
64
  - num_epochs: 3
65
 
 
67
 
68
  | Training Loss | Epoch | Step | Validation Loss | Exact Match | F1 |
69
  |:-------------:|:------:|:-----:|:---------------:|:-----------:|:-------:|
70
+ | 3.0349 | 0.0335 | 200 | 2.7893 | 0.0 | 0.0067 |
71
+ | 2.3253 | 0.0670 | 400 | 2.1518 | 11.7680 | 15.5697 |
72
+ | 2.0108 | 0.1006 | 600 | 2.0181 | 19.2455 | 23.7003 |
73
+ | 1.9105 | 0.1341 | 800 | 1.8422 | 24.4161 | 28.9644 |
74
+ | 1.893 | 0.1676 | 1000 | 1.7602 | 29.6921 | 35.4185 |
75
+ | 1.7635 | 0.2011 | 1200 | 1.7062 | 26.7003 | 31.6106 |
76
+ | 1.8434 | 0.2347 | 1400 | 1.6456 | 31.7693 | 38.0953 |
77
+ | 1.6387 | 0.2682 | 1600 | 1.6191 | 29.2502 | 35.6592 |
78
+ | 1.6512 | 0.3017 | 1800 | 1.5874 | 36.6594 | 44.5029 |
79
+ | 1.6318 | 0.3352 | 2000 | 1.5478 | 31.1712 | 37.5434 |
80
+ | 1.6269 | 0.3688 | 2200 | 1.5439 | 37.7275 | 45.9815 |
81
+ | 1.5866 | 0.4023 | 2400 | 1.5259 | 33.2852 | 40.1296 |
82
+ | 1.5102 | 0.4358 | 2600 | 1.5545 | 31.8182 | 38.3162 |
83
+ | 1.5253 | 0.4693 | 2800 | 1.4899 | 41.2113 | 50.4395 |
84
+ | 1.4366 | 0.5028 | 3000 | 1.4812 | 40.2321 | 49.6351 |
85
+ | 1.6307 | 0.5364 | 3200 | 1.4455 | 41.1860 | 49.6116 |
86
+ | 1.4605 | 0.5699 | 3400 | 1.4304 | 38.4629 | 46.3922 |
87
+ | 1.4125 | 0.6034 | 3600 | 1.4257 | 41.0046 | 50.8304 |
88
+ | 1.4126 | 0.6369 | 3800 | 1.4215 | 41.3979 | 50.7890 |
89
+ | 1.5035 | 0.6705 | 4000 | 1.3847 | 39.6329 | 48.5817 |
90
+ | 1.3627 | 0.7040 | 4200 | 1.4561 | 29.0115 | 34.8629 |
91
+ | 1.4172 | 0.7375 | 4400 | 1.3951 | 45.1590 | 55.5680 |
92
+ | 1.4262 | 0.7710 | 4600 | 1.3571 | 42.7241 | 51.8206 |
93
+ | 1.3756 | 0.8046 | 4800 | 1.3717 | 43.1109 | 51.3852 |
94
+ | 1.3978 | 0.8381 | 5000 | 1.4136 | 48.0715 | 59.8789 |
95
+ | 1.4521 | 0.8716 | 5200 | 1.3389 | 41.3291 | 50.7222 |
96
+ | 1.4738 | 0.9051 | 5400 | 1.3281 | 38.1464 | 45.8767 |
97
+ | 1.372 | 0.9387 | 5600 | 1.3212 | 44.6938 | 54.1932 |
98
+ | 1.414 | 0.9722 | 5800 | 1.3104 | 45.1054 | 55.2289 |
99
+ | 1.3008 | 1.0057 | 6000 | 1.3411 | 45.8649 | 56.2610 |
100
+ | 1.0646 | 1.0392 | 6200 | 1.4034 | 39.6067 | 47.5529 |
101
+ | 1.0405 | 1.0727 | 6400 | 1.4081 | 42.7331 | 51.7438 |
102
+ | 1.0141 | 1.1063 | 6600 | 1.4326 | 40.6200 | 49.2831 |
103
+ | 1.1305 | 1.1398 | 6800 | 1.3429 | 46.5557 | 56.9270 |
104
+ | 1.0131 | 1.1733 | 7000 | 1.3695 | 48.7474 | 60.3360 |
105
+ | 1.1332 | 1.2068 | 7200 | 1.3221 | 44.8748 | 54.8693 |
106
+ | 1.1572 | 1.2404 | 7400 | 1.3601 | 49.7453 | 60.7304 |
107
+ | 1.0497 | 1.2739 | 7600 | 1.3221 | 48.4678 | 59.5859 |
108
+ | 1.1202 | 1.3074 | 7800 | 1.2960 | 42.6078 | 52.2938 |
109
+ | 1.1005 | 1.3409 | 8000 | 1.3422 | 49.1114 | 60.8679 |
110
+ | 1.0976 | 1.3745 | 8200 | 1.3270 | 46.8241 | 57.5165 |
111
+ | 1.1028 | 1.4080 | 8400 | 1.2932 | 45.9230 | 57.3813 |
112
+ | 0.9758 | 1.4415 | 8600 | 1.3032 | 45.1296 | 55.6205 |
113
+ | 1.0391 | 1.4750 | 8800 | 1.2878 | 48.0178 | 58.6035 |
114
+ | 1.1021 | 1.5085 | 9000 | 1.2840 | 48.8204 | 59.6174 |
115
+ | 1.0591 | 1.5421 | 9200 | 1.3227 | 46.5811 | 57.0738 |
116
+ | 1.0742 | 1.5756 | 9400 | 1.2771 | 44.4915 | 54.2228 |
117
+ | 1.1314 | 1.6091 | 9600 | 1.3067 | 49.2240 | 60.6819 |
118
+ | 1.0721 | 1.6426 | 9800 | 1.2839 | 46.6994 | 57.4786 |
119
+ | 1.1123 | 1.6762 | 10000 | 1.2718 | 47.7972 | 59.1149 |
120
+ | 1.0766 | 1.7097 | 10200 | 1.2688 | 49.3350 | 61.0489 |
121
+ | 1.1244 | 1.7432 | 10400 | 1.2575 | 48.4543 | 59.6361 |
122
+ | 1.0744 | 1.7767 | 10600 | 1.2788 | 48.7775 | 59.4327 |
123
+ | 1.0186 | 1.8103 | 10800 | 1.2620 | 48.6458 | 59.9898 |
124
+ | 0.9617 | 1.8438 | 11000 | 1.3137 | 43.1942 | 52.7838 |
125
+ | 0.9996 | 1.8773 | 11200 | 1.2786 | 50.3568 | 62.0152 |
126
+ | 0.9281 | 1.9108 | 11400 | 1.2849 | 46.7113 | 56.7769 |
127
+ | 1.0331 | 1.9444 | 11600 | 1.2693 | 46.9996 | 57.3083 |
128
+ | 1.0482 | 1.9779 | 11800 | 1.2636 | 44.8373 | 54.6672 |
129
+ | 0.7695 | 2.0114 | 12000 | 1.3635 | 45.9601 | 56.4656 |
130
+ | 0.7887 | 2.0449 | 12200 | 1.4005 | 48.8684 | 60.5211 |
131
+ | 0.782 | 2.0784 | 12400 | 1.3826 | 49.2449 | 59.9969 |
132
+ | 0.7674 | 2.1120 | 12600 | 1.3707 | 47.4254 | 58.1781 |
133
+ | 0.7597 | 2.1455 | 12800 | 1.3924 | 48.4130 | 59.9062 |
134
+ | 0.7555 | 2.1790 | 13000 | 1.3777 | 47.3922 | 58.6007 |
135
+ | 0.7261 | 2.2125 | 13200 | 1.4037 | 50.1306 | 61.4821 |
136
+ | 0.7681 | 2.2461 | 13400 | 1.4149 | 48.0112 | 59.3190 |
137
+ | 0.7899 | 2.2796 | 13600 | 1.3700 | 46.4242 | 56.9562 |
138
 
139
 
140
  ### Framework versions
 
142
  - Transformers 4.48.0.dev0
143
  - Pytorch 2.4.1+cu121
144
  - Datasets 2.20.0
145
+ - Tokenizers 0.21.0
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a5e8f06fcfe865348466eaa2bb2ad4f060878edc19510b9bd8cfe35edbdc4189
3
  size 440136504
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cb9fe2a794a6c2eaee5733751b15ef8674fffab69b65b463a9a17d804e80bcb6
3
  size 440136504
tokenizer.json CHANGED
@@ -1,21 +1,7 @@
1
  {
2
  "version": "1.0",
3
- "truncation": {
4
- "direction": "Right",
5
- "max_length": 384,
6
- "strategy": "OnlySecond",
7
- "stride": 128
8
- },
9
- "padding": {
10
- "strategy": {
11
- "Fixed": 384
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,
 
1
  {
2
  "version": "1.0",
3
+ "truncation": null,
4
+ "padding": null,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5
  "added_tokens": [
6
  {
7
  "id": 0,
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:dba1cec0df5b83a5dd64705bc683709ebbb2f91d1002f087d496b4caf03ba059
3
  size 5432
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:db631ee58955f03d65737d8a3e92912477994ac879715d010f031d1436b46785
3
  size 5432