Hugging Face's logo

santis2
/
distilgpt2-alpaca-instruction-fine-tuning-qlora

Generated from Trainer
Model card Files Community
santis2 commited on
Commit
55e50b3
•
1 Parent(s): 323c5d3

End of training

Browse files
Files changed (3) hide show
  1. README.md +15 -66
  2. adapter_model.bin +1 -1
  3. training_args.bin +1 -1
README.md CHANGED
@@ -15,7 +15,7 @@ should probably proofread and complete it, then remove this comment. -->
15
 
16
  This model is a fine-tuned version of [distilgpt2](https://huggingface.co/distilgpt2) on an unknown dataset.
17
  It achieves the following results on the evaluation set:
18
- - Loss: 2.6367
19
 
20
  ## Model description
21
 
@@ -35,78 +35,27 @@ More information needed
35
 
36
  The following hyperparameters were used during training:
37
  - learning_rate: 0.0005
38
- - train_batch_size: 24
39
- - eval_batch_size: 24
40
  - seed: 42
41
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
42
  - lr_scheduler_type: cosine
43
  - lr_scheduler_warmup_steps: 1000
44
- - num_epochs: 40
45
 
46
  ### Training results
47
 
48
- | Training Loss | Epoch | Step | Validation Loss |
49
- |:-------------:|:-----:|:-----:|:---------------:|
50
- | 3.6682 | 0.66 | 1000 | 2.7578 |
51
- | 3.0752 | 1.32 | 2000 | 2.6973 |
52
- | 3.0293 | 1.99 | 3000 | 2.7012 |
53
- | 3.0137 | 2.65 | 4000 | 2.6465 |
54
- | 3.0057 | 3.31 | 5000 | 2.6562 |
55
- | 3.0113 | 3.97 | 6000 | 2.6621 |
56
- | 3.0044 | 4.64 | 7000 | 2.6426 |
57
- | 3.0059 | 5.3 | 8000 | 2.6719 |
58
- | 3.0087 | 5.96 | 9000 | 2.6602 |
59
- | 3.0002 | 6.62 | 10000 | 2.6406 |
60
- | 2.9971 | 7.28 | 11000 | 2.6680 |
61
- | 2.9896 | 7.95 | 12000 | 2.6602 |
62
- | 2.9936 | 8.61 | 13000 | 2.6699 |
63
- | 2.9985 | 9.27 | 14000 | 2.6641 |
64
- | 2.9845 | 9.93 | 15000 | 2.6660 |
65
- | 2.9953 | 10.6 | 16000 | 2.6523 |
66
- | 2.9858 | 11.26 | 17000 | 2.6621 |
67
- | 2.9892 | 11.92 | 18000 | 2.6445 |
68
- | 2.9869 | 12.58 | 19000 | 2.625 |
69
- | 2.9899 | 13.25 | 20000 | 2.6465 |
70
- | 2.9824 | 13.91 | 21000 | 2.6836 |
71
- | 2.9824 | 14.57 | 22000 | 2.6445 |
72
- | 2.9869 | 15.23 | 23000 | 2.6641 |
73
- | 2.9864 | 15.89 | 24000 | 2.6543 |
74
- | 2.9815 | 16.56 | 25000 | 2.6309 |
75
- | 2.9806 | 17.22 | 26000 | 2.6777 |
76
- | 2.9884 | 17.88 | 27000 | 2.6270 |
77
- | 2.9919 | 18.54 | 28000 | 2.6445 |
78
- | 2.9842 | 19.21 | 29000 | 2.6758 |
79
- | 2.9771 | 19.87 | 30000 | 2.6543 |
80
- | 2.9824 | 20.53 | 31000 | 2.6523 |
81
- | 2.9889 | 21.19 | 32000 | 2.6523 |
82
- | 2.9816 | 21.85 | 33000 | 2.6270 |
83
- | 2.9844 | 22.52 | 34000 | 2.6562 |
84
- | 2.9781 | 23.18 | 35000 | 2.6328 |
85
- | 2.9802 | 23.84 | 36000 | 2.6289 |
86
- | 2.9783 | 24.5 | 37000 | 2.6270 |
87
- | 2.9832 | 25.17 | 38000 | 2.6406 |
88
- | 2.9759 | 25.83 | 39000 | 2.6289 |
89
- | 2.9805 | 26.49 | 40000 | 2.6387 |
90
- | 2.9795 | 27.15 | 41000 | 2.6367 |
91
- | 2.9728 | 27.81 | 42000 | 2.6367 |
92
- | 2.978 | 28.48 | 43000 | 2.6426 |
93
- | 2.9779 | 29.14 | 44000 | 2.6348 |
94
- | 2.977 | 29.8 | 45000 | 2.6406 |
95
- | 2.9692 | 30.46 | 46000 | 2.6348 |
96
- | 2.9815 | 31.13 | 47000 | 2.6406 |
97
- | 2.9706 | 31.79 | 48000 | 2.6387 |
98
- | 2.9764 | 32.45 | 49000 | 2.6367 |
99
- | 2.9733 | 33.11 | 50000 | 2.6230 |
100
- | 2.9751 | 33.77 | 51000 | 2.6504 |
101
- | 2.968 | 34.44 | 52000 | 2.6465 |
102
- | 2.972 | 35.1 | 53000 | 2.6523 |
103
- | 2.9744 | 35.76 | 54000 | 2.6406 |
104
- | 2.9611 | 36.42 | 55000 | 2.6328 |
105
- | 2.9786 | 37.09 | 56000 | 2.6348 |
106
- | 2.9712 | 37.75 | 57000 | 2.6406 |
107
- | 2.9722 | 38.41 | 58000 | 2.6348 |
108
- | 2.9709 | 39.07 | 59000 | 2.6367 |
109
- | 2.9735 | 39.74 | 60000 | 2.6367 |
110
 
111
 
112
  ### Framework versions
 
15
 
16
  This model is a fine-tuned version of [distilgpt2](https://huggingface.co/distilgpt2) on an unknown dataset.
17
  It achieves the following results on the evaluation set:
18
+ - Loss: 2.2461
19
 
20
  ## Model description
21
 
 
35
 
36
  The following hyperparameters were used during training:
37
  - learning_rate: 0.0005
38
+ - train_batch_size: 4
39
+ - eval_batch_size: 4
40
  - seed: 42
41
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
42
  - lr_scheduler_type: cosine
43
  - lr_scheduler_warmup_steps: 1000
44
+ - num_epochs: 1
45
 
46
  ### Training results
47
 
48
+ | Training Loss | Epoch | Step | Validation Loss |
49
+ |:-------------:|:-----:|:----:|:---------------:|
50
+ | 2.9811 | 0.11 | 1000 | 2.4141 |
51
+ | 2.5312 | 0.22 | 2000 | 2.3164 |
52
+ | 2.4908 | 0.33 | 3000 | 2.2871 |
53
+ | 2.4785 | 0.44 | 4000 | 2.2754 |
54
+ | 2.4518 | 0.55 | 5000 | 2.2832 |
55
+ | 2.4277 | 0.66 | 6000 | 2.2578 |
56
+ | 2.4352 | 0.77 | 7000 | 2.25 |
57
+ | 2.4171 | 0.88 | 8000 | 2.2480 |
58
+ | 2.4138 | 0.99 | 9000 | 2.2461 |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
59
 
60
 
61
  ### Framework versions
adapter_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e22809645a4aafe8f25ef37433cb46d18e7bed28a4b2ca946e7d93ed00b1e49e
3
  size 418013
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0bd87c89fcd8fe6d7f4864e45015714123e4b18094325f9796b81192c55eec2a
3
  size 418013
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d92186ffc4adc2ba0962f503328a317336dcea34dd6748736f99e51aaac2c7cb
3
  size 4091
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e80b6fef053a79b4d6a8998a5ba2c4f54d311285bd63be90fb5dd6dc44493072
3
  size 4091