Hugging Face's logo

distily
/
short_gpt2

TensorBoard
Safetensors
Distily
gpt2
Generated from Trainer
Model card Files Metrics Community
lapp0 commited on
Commit
c32de4f
·
verified ·
1 Parent(s): 22c1228

End of training

Browse files
Files changed (14) hide show
  1. README.md +135 -0
  2. config.json +46 -0
  3. generation_config.json +6 -0
  4. logs/events.out.tfevents.1724126411.f383272e719b +3 -0
  5. logs/events.out.tfevents.1724126630.f383272e719b +3 -0
  6. logs/events.out.tfevents.1724128308.f383272e719b +3 -0
  7. logs/events.out.tfevents.1724139617.f383272e719b +3 -0
  8. merges.txt +0 -0
  9. model.safetensors +3 -0
  10. special_tokens_map.json +6 -0
  11. tokenizer.json +0 -0
  12. tokenizer_config.json +20 -0
  13. training_args.bin +3 -0
  14. vocab.json +0 -0
README.md ADDED
@@ -0,0 +1,135 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: distilbert/distilgpt2
3
+ library_name: Distily
4
+ license: apache-2.0
5
+ tags:
6
+ - generated_from_trainer
7
+ model-index:
8
+ - name: short_gpt2
9
+ results: []
10
+ ---
11
+
12
+ # short_gpt2
13
+
14
+ This student model is distilled from the teacher model [gpt2](https://huggingface.co/gpt2) using the dataset (unspecified).
15
+
16
+ The [Distily](https://github.com/lapp0/distily) library was used for this distillation.
17
+
18
+ It achieves the following results on the evaluation set:
19
+ - eval_enwikippl: 96.0
20
+ - eval_frwikippl: 372.0
21
+ - eval_zhwikippl: 128.0
22
+ - eval_tinystoriesppl: 74.5
23
+ - eval_loss: 0.7694
24
+ - eval_runtime: 25.5363
25
+ - eval_samples_per_second: 97.9
26
+ - eval_steps_per_second: 12.257
27
+
28
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
29
+ should probably proofread and complete it, then remove this comment.
30
+
31
+ ## Model description
32
+
33
+ More information needed
34
+
35
+ ## Intended uses & limitations
36
+
37
+ More information needed
38
+
39
+ ## Training and evaluation data
40
+
41
+ More information needed
42
+ -->
43
+
44
+ ## Training procedure
45
+
46
+ ### Training hyperparameters
47
+
48
+ The following hyperparameters were used during training:
49
+ - distillation_objective: DistillationObjective(logits_loss_component=LossComponent(label=logits, weight=1, loss_fn=kl, layer_mapper=None, projector=None), hs_loss_component=LossComponent(label=hs, weight=0, loss_fn=None, layer_mapper=None, projector=None), attn_loss_component=LossComponent(label=attn, weight=0, loss_fn=None, layer_mapper=None, projector=None))
50
+ - train_embeddings: True
51
+ - learning_rate: 0.0001
52
+ - train_batch_size: 4
53
+ - eval_batch_size: 8
54
+ - seed: 42
55
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
56
+ - lr_scheduler_type: constant
57
+ - lr_scheduler_warmup_ratio: 0.2
58
+ - num_epochs: 1.0
59
+
60
+ ### Resource Usage
61
+ Peak GPU Memory: 7.2012 GB
62
+
63
+ ### Eval-Phase Metrics
64
+ | step | epoch | enwikippl | frwikippl | loss | runtime | samples_per_second | steps_per_second | tinystoriesppl | zhwikippl |
65
+ | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
66
+ | **teacher eval** | | 43.25 | 61.25 | | | | | 11.6875 | 19.125 |
67
+ | 0 | 0 | 1752346656768.0 | 132491151147008.0 | 20.4918 | 25.5082 | 98.008 | 12.271 | 5335154688.0 | 43705587204096.0 |
68
+ | 1000 | 0.0162 | 274.0 | 1200.0 | 1.4194 | 25.4948 | 98.059 | 12.277 | 220.0 | 184.0 |
69
+ | 2000 | 0.0323 | 200.0 | 680.0 | 1.2412 | 25.5013 | 98.034 | 12.274 | 163.0 | 151.0 |
70
+ | 3000 | 0.0485 | 165.0 | 644.0 | 1.1185 | 25.5706 | 97.769 | 12.241 | 138.0 | 149.0 |
71
+ | 4000 | 0.0646 | 146.0 | 576.0 | 1.0289 | 25.4758 | 98.132 | 12.286 | 114.5 | 134.0 |
72
+ | 5000 | 0.0808 | 131.0 | 520.0 | 0.9689 | 25.4905 | 98.076 | 12.279 | 100.5 | 139.0 |
73
+ | 6000 | 0.0970 | 117.0 | 456.0 | 0.9014 | 25.4949 | 98.059 | 12.277 | 92.5 | 140.0 |
74
+ | 7000 | 0.1131 | 109.5 | 412.0 | 0.8654 | 25.5404 | 97.884 | 12.255 | 88.0 | 148.0 |
75
+ | 8000 | 0.1293 | 104.0 | 414.0 | 0.8141 | 25.5188 | 97.967 | 12.265 | 80.0 | 131.0 |
76
+ | 9000 | 0.1455 | 96.0 | 372.0 | 0.7694 | 25.5363 | 97.9 | 12.257 | 74.5 | 128.0 |
77
+ | 10000 | 0.1616 | 90.0 | 372.0 | 0.7269 | 25.4962 | 98.054 | 12.276 | 71.5 | 125.0 |
78
+ | 11000 | 0.1778 | 88.0 | 336.0 | 0.6943 | 25.4752 | 98.135 | 12.286 | 69.5 | 122.0 |
79
+ | 12000 | 0.1939 | 84.5 | 336.0 | 0.6694 | 25.4772 | 98.127 | 12.285 | 66.5 | 134.0 |
80
+ | 13000 | 0.2101 | 79.5 | 284.0 | 0.6458 | 25.4687 | 98.16 | 12.29 | 65.0 | 150.0 |
81
+ | 14000 | 0.2263 | 78.5 | 298.0 | 0.6230 | 25.5385 | 97.891 | 12.256 | 62.75 | 150.0 |
82
+ | 15000 | 0.2424 | 75.5 | 272.0 | 0.6116 | 25.5112 | 97.996 | 12.269 | 59.75 | 135.0 |
83
+ | 16000 | 0.2586 | 76.0 | 260.0 | 0.6013 | 25.5149 | 97.982 | 12.267 | 59.0 | 162.0 |
84
+ | 17000 | 0.2747 | 75.0 | 284.0 | 0.5878 | 25.5099 | 98.001 | 12.27 | 60.75 | 114.5 |
85
+ | 18000 | 0.2909 | 74.5 | 260.0 | 0.5739 | 25.5029 | 98.028 | 12.273 | 60.75 | 141.0 |
86
+ | 19000 | 0.3071 | 72.0 | 282.0 | 0.5736 | 25.5176 | 97.972 | 12.266 | 57.5 | 135.0 |
87
+ | 20000 | 0.3232 | 72.0 | 264.0 | 0.5581 | 25.5059 | 98.016 | 12.272 | 55.5 | 129.0 |
88
+ | 21000 | 0.3394 | 72.0 | 262.0 | 0.5565 | 25.5747 | 97.753 | 12.239 | 56.25 | 136.0 |
89
+ | 22000 | 0.3556 | 73.5 | 252.0 | 0.5476 | 25.5063 | 98.015 | 12.271 | 54.25 | 116.5 |
90
+ | 23000 | 0.3717 | 71.5 | 252.0 | 0.5455 | 25.4614 | 98.188 | 12.293 | 52.0 | 124.5 |
91
+ | 24000 | 0.3879 | 72.0 | 260.0 | 0.5401 | 25.4593 | 98.196 | 12.294 | 53.5 | 136.0 |
92
+ | 25000 | 0.4040 | 71.0 | 244.0 | 0.5351 | 25.4834 | 98.103 | 12.283 | 53.25 | 107.5 |
93
+ | 26000 | 0.4202 | 68.5 | 227.0 | 0.5321 | 25.4557 | 98.21 | 12.296 | 51.25 | 94.0 |
94
+ | 27000 | 0.4364 | 73.0 | 235.0 | 0.5286 | 25.4824 | 98.107 | 12.283 | 51.75 | 115.5 |
95
+ | 28000 | 0.4525 | 70.0 | 248.0 | 0.5310 | 25.5236 | 97.949 | 12.263 | 50.5 | 117.5 |
96
+ | 29000 | 0.4687 | 70.5 | 241.0 | 0.5233 | 25.4617 | 98.187 | 12.293 | 49.25 | 171.0 |
97
+ | 30000 | 0.4848 | 70.0 | 228.0 | 0.5170 | 25.5019 | 98.032 | 12.274 | 54.0 | 106.0 |
98
+ | 31000 | 0.5010 | 69.0 | 253.0 | 0.5187 | 25.5293 | 97.927 | 12.26 | 51.5 | 113.0 |
99
+ | 32000 | 0.5172 | 73.5 | 243.0 | 0.5160 | 25.5033 | 98.027 | 12.273 | 53.5 | 85.0 |
100
+ | 33000 | 0.5333 | 75.0 | 264.0 | 0.5181 | 25.516 | 97.978 | 12.267 | 52.5 | 87.5 |
101
+ | 34000 | 0.5495 | 71.0 | 228.0 | 0.5141 | 25.4586 | 98.198 | 12.294 | 51.75 | 128.0 |
102
+ | 35000 | 0.5657 | 69.5 | 241.0 | 0.5159 | 25.4876 | 98.087 | 12.28 | 52.0 | 142.0 |
103
+ | 36000 | 0.5818 | 69.0 | 254.0 | 0.5107 | 25.497 | 98.051 | 12.276 | 52.75 | 111.0 |
104
+ | 37000 | 0.5980 | 69.0 | 225.0 | 0.5010 | 25.5451 | 97.866 | 12.253 | 52.5 | 145.0 |
105
+ | 38000 | 0.6141 | 70.0 | 220.0 | 0.5100 | 25.529 | 97.928 | 12.261 | 51.5 | 516.0 |
106
+ | 39000 | 0.6303 | 67.0 | 235.0 | 0.5033 | 25.495 | 98.058 | 12.277 | 51.0 | 102.0 |
107
+ | 40000 | 0.6465 | 67.5 | 231.0 | 0.5040 | 25.4679 | 98.163 | 12.29 | 48.5 | 95.0 |
108
+ | 41000 | 0.6626 | 68.5 | 217.0 | 0.4893 | 25.4787 | 98.121 | 12.285 | 51.25 | 92.5 |
109
+ | 42000 | 0.6788 | 68.5 | 211.0 | 0.4965 | 25.4674 | 98.165 | 12.29 | 50.5 | 143.0 |
110
+ | 43000 | 0.6949 | 68.0 | 220.0 | 0.4998 | 25.4714 | 98.149 | 12.288 | 50.25 | 120.0 |
111
+ | 44000 | 0.7111 | 69.5 | 224.0 | 0.4985 | 25.527 | 97.936 | 12.262 | 51.0 | 109.5 |
112
+ | 45000 | 0.7273 | 69.0 | 230.0 | 0.5052 | 25.4809 | 98.113 | 12.284 | 50.75 | 109.5 |
113
+ | 46000 | 0.7434 | 67.5 | 221.0 | 0.4921 | 25.4876 | 98.087 | 12.28 | 49.25 | 92.0 |
114
+ | 47000 | 0.7596 | 69.0 | 230.0 | 0.4942 | 25.4768 | 98.128 | 12.286 | 49.5 | 164.0 |
115
+ | 48000 | 0.7758 | 67.5 | 219.0 | 0.4892 | 25.4923 | 98.069 | 12.278 | 52.25 | 88.0 |
116
+ | 49000 | 0.7919 | 70.0 | 247.0 | 0.4904 | 25.485 | 98.097 | 12.282 | 49.0 | 126.5 |
117
+ | 50000 | 0.8081 | 66.0 | 219.0 | 0.4852 | 25.4685 | 98.161 | 12.29 | 48.0 | 120.0 |
118
+ | 51000 | 0.8242 | 67.5 | 230.0 | 0.4926 | 25.5059 | 98.017 | 12.272 | 47.75 | 99.0 |
119
+ | 52000 | 0.8404 | 67.5 | 224.0 | 0.4838 | 25.5222 | 97.954 | 12.264 | 50.25 | 144.0 |
120
+ | 53000 | 0.8566 | 67.5 | 213.0 | 0.4853 | 25.5206 | 97.96 | 12.265 | 48.75 | 114.0 |
121
+ | 54000 | 0.8727 | 64.5 | 228.0 | 0.4889 | 25.494 | 98.062 | 12.277 | 49.5 | 116.0 |
122
+ | 55000 | 0.8889 | 65.5 | 231.0 | 0.4819 | 25.5118 | 97.994 | 12.269 | 49.25 | 113.5 |
123
+ | 56000 | 0.9051 | 67.0 | 223.0 | 0.4872 | 25.4843 | 98.1 | 12.282 | 48.0 | 104.0 |
124
+ | 57000 | 0.9212 | 68.5 | 234.0 | 0.4875 | 25.4685 | 98.16 | 12.29 | 49.75 | 131.0 |
125
+ | 58000 | 0.9374 | 65.0 | 222.0 | 0.4854 | 25.4413 | 98.266 | 12.303 | 46.75 | 132.0 |
126
+ | 59000 | 0.9535 | 65.5 | 211.0 | 0.4775 | 25.4688 | 98.159 | 12.29 | 49.5 | 132.0 |
127
+ | 60000 | 0.9697 | 67.0 | 221.0 | 0.4806 | 25.4434 | 98.257 | 12.302 | 49.0 | 106.5 |
128
+ | 61000 | 0.9859 | 65.5 | 216.0 | 0.4853 | 25.4536 | 98.218 | 12.297 | 50.25 | 109.5 |
129
+ | 61875 | 1.0 | 62.75 | 222.0 | 0.4830 | 25.7335 | 97.15 | 12.163 | 49.25 | 204.0 |
130
+
131
+ ### Framework versions
132
+ - Distily 0.2.0
133
+ - Transformers 4.44.0
134
+ - Pytorch 2.3.0
135
+ - Datasets 2.21.0
config.json ADDED
@@ -0,0 +1,46 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "distilbert/distilgpt2",
3
+ "_num_labels": 1,
4
+ "activation_function": "gelu_new",
5
+ "architectures": [
6
+ "GPT2LMHeadModel"
7
+ ],
8
+ "attn_pdrop": 0.1,
9
+ "bos_token_id": 50256,
10
+ "embd_pdrop": 0.1,
11
+ "eos_token_id": 50256,
12
+ "id2label": {
13
+ "0": "LABEL_0"
14
+ },
15
+ "initializer_range": 0.02,
16
+ "label2id": {
17
+ "LABEL_0": 0
18
+ },
19
+ "layer_norm_epsilon": 1e-05,
20
+ "model_type": "gpt2",
21
+ "n_ctx": 1024,
22
+ "n_embd": 768,
23
+ "n_head": 12,
24
+ "n_inner": null,
25
+ "n_layer": 6,
26
+ "n_positions": 1024,
27
+ "reorder_and_upcast_attn": false,
28
+ "resid_pdrop": 0.1,
29
+ "scale_attn_by_inverse_layer_idx": false,
30
+ "scale_attn_weights": true,
31
+ "summary_activation": null,
32
+ "summary_first_dropout": 0.1,
33
+ "summary_proj_to_labels": true,
34
+ "summary_type": "cls_index",
35
+ "summary_use_proj": true,
36
+ "task_specific_params": {
37
+ "text-generation": {
38
+ "do_sample": true,
39
+ "max_length": 50
40
+ }
41
+ },
42
+ "torch_dtype": "bfloat16",
43
+ "transformers_version": "4.44.0",
44
+ "use_cache": true,
45
+ "vocab_size": 50257
46
+ }
generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 50256,
4
+ "eos_token_id": 50256,
5
+ "transformers_version": "4.44.0"
6
+ }
logs/events.out.tfevents.1724126411.f383272e719b ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:203a63c9e3ab4454c54a8449b8cf08e30777cd6c2efe1f12c3461e7a5fd9299c
3
+ size 5255
logs/events.out.tfevents.1724126630.f383272e719b ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d5638a5baf8421a50b6512235461668012f2061facc3fbbd03542e22a5635b5e
3
+ size 3784372
logs/events.out.tfevents.1724128308.f383272e719b ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:72a45a0fe63bf2e503dfed4c816f8bc03dcf04e6f1781fe40082cdbcfdab7579
3
+ size 29652657
logs/events.out.tfevents.1724139617.f383272e719b ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:63cb357b87d90af8da2622792c97b046bc9770a0e3614b7f824f89fed8bba3cb
3
+ size 312
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4ba7f2cb76ed1a3b42486c9431bdc0e8fc1d7034a34380fa32c0df3efb041d6b
3
+ size 163832792
special_tokens_map.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<|endoftext|>",
3
+ "eos_token": "<|endoftext|>",
4
+ "pad_token": "<|endoftext|>",
5
+ "unk_token": "<|endoftext|>"
6
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "added_tokens_decoder": {
4
+ "50256": {
5
+ "content": "<|endoftext|>",
6
+ "lstrip": false,
7
+ "normalized": true,
8
+ "rstrip": false,
9
+ "single_word": false,
10
+ "special": true
11
+ }
12
+ },
13
+ "bos_token": "<|endoftext|>",
14
+ "clean_up_tokenization_spaces": true,
15
+ "eos_token": "<|endoftext|>",
16
+ "model_max_length": 1024,
17
+ "pad_token": "<|endoftext|>",
18
+ "tokenizer_class": "GPT2Tokenizer",
19
+ "unk_token": "<|endoftext|>"
20
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1d79ceb7e1da484c1681c46025ad4bc1e4ef1adc49c82179b48756fc2a218da0
3
+ size 1017899016
vocab.json ADDED
The diff for this file is too large to render. See raw diff