Hugging Face's logo

Gunslinger3D
/
fine-tuning-Phi2-with-webglm-qa-with-lora

PEFT
Safetensors
Generated from Trainer
Model card Files
xet
Community
Gunslinger3D commited on
Commit
5539dc8
·
verified ·
1 Parent(s): 05ed7d7

fine-tuning-Phi2-with-webglm-qa-with-lora

Browse files
Files changed (4) hide show
  1. README.md +53 -33
  2. adapter_config.json +3 -3
  3. adapter_model.safetensors +1 -1
  4. training_args.bin +1 -1
README.md CHANGED
@@ -16,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
16
 
17
  This model is a fine-tuned version of [microsoft/phi-2](https://huggingface.co/microsoft/phi-2) on an unknown dataset.
18
  It achieves the following results on the evaluation set:
19
- - Loss: 0.1475
20
 
21
  ## Model description
22
 
@@ -43,44 +43,64 @@ The following hyperparameters were used during training:
43
  - total_train_batch_size: 10
44
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
45
  - lr_scheduler_type: linear
46
- - lr_scheduler_warmup_steps: 30
47
- - training_steps: 300
48
  - mixed_precision_training: Native AMP
49
 
50
  ### Training results
51
 
52
  | Training Loss | Epoch | Step | Validation Loss |
53
  |:-------------:|:-----:|:----:|:---------------:|
54
- | No log | 0.2 | 10 | 7.7121 |
55
- | 7.4808 | 0.4 | 20 | 4.3398 |
56
- | 7.4808 | 0.6 | 30 | 0.6362 |
57
- | 1.5296 | 0.8 | 40 | 0.5285 |
58
- | 1.5296 | 1.0 | 50 | 0.4668 |
59
- | 0.3883 | 1.2 | 60 | 0.4194 |
60
- | 0.3883 | 1.39 | 70 | 0.3737 |
61
- | 0.3482 | 1.59 | 80 | 0.3338 |
62
- | 0.3482 | 1.79 | 90 | 0.3036 |
63
- | 0.2296 | 1.99 | 100 | 0.2802 |
64
- | 0.2296 | 2.19 | 110 | 0.2595 |
65
- | 0.212 | 2.39 | 120 | 0.2452 |
66
- | 0.212 | 2.59 | 130 | 0.2307 |
67
- | 0.1943 | 2.79 | 140 | 0.2145 |
68
- | 0.1943 | 2.99 | 150 | 0.2031 |
69
- | 0.1635 | 3.19 | 160 | 0.1957 |
70
- | 0.1635 | 3.39 | 170 | 0.1857 |
71
- | 0.1543 | 3.59 | 180 | 0.1788 |
72
- | 0.1543 | 3.78 | 190 | 0.1732 |
73
- | 0.1492 | 3.98 | 200 | 0.1687 |
74
- | 0.1492 | 4.18 | 210 | 0.1650 |
75
- | 0.1327 | 4.38 | 220 | 0.1632 |
76
- | 0.1327 | 4.58 | 230 | 0.1597 |
77
- | 0.1359 | 4.78 | 240 | 0.1552 |
78
- | 0.1359 | 4.98 | 250 | 0.1522 |
79
- | 0.1367 | 5.18 | 260 | 0.1506 |
80
- | 0.1367 | 5.38 | 270 | 0.1495 |
81
- | 0.1204 | 5.58 | 280 | 0.1484 |
82
- | 0.1204 | 5.78 | 290 | 0.1477 |
83
- | 0.125 | 5.98 | 300 | 0.1475 |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
84
 
85
 
86
  ### Framework versions
 
16
 
17
  This model is a fine-tuned version of [microsoft/phi-2](https://huggingface.co/microsoft/phi-2) on an unknown dataset.
18
  It achieves the following results on the evaluation set:
19
+ - Loss: 0.1008
20
 
21
  ## Model description
22
 
 
43
  - total_train_batch_size: 10
44
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
45
  - lr_scheduler_type: linear
46
+ - lr_scheduler_warmup_steps: 50
47
+ - training_steps: 500
48
  - mixed_precision_training: Native AMP
49
 
50
  ### Training results
51
 
52
  | Training Loss | Epoch | Step | Validation Loss |
53
  |:-------------:|:-----:|:----:|:---------------:|
54
+ | No log | 0.2 | 10 | 7.9734 |
55
+ | No log | 0.4 | 20 | 6.2577 |
56
+ | No log | 0.6 | 30 | 2.8966 |
57
+ | No log | 0.8 | 40 | 0.5650 |
58
+ | 4.4766 | 1.0 | 50 | 0.5106 |
59
+ | 4.4766 | 1.2 | 60 | 0.4513 |
60
+ | 4.4766 | 1.39 | 70 | 0.4008 |
61
+ | 4.4766 | 1.59 | 80 | 0.3539 |
62
+ | 4.4766 | 1.79 | 90 | 0.3156 |
63
+ | 0.3251 | 1.99 | 100 | 0.2877 |
64
+ | 0.3251 | 2.19 | 110 | 0.2631 |
65
+ | 0.3251 | 2.39 | 120 | 0.2464 |
66
+ | 0.3251 | 2.59 | 130 | 0.2303 |
67
+ | 0.3251 | 2.79 | 140 | 0.2117 |
68
+ | 0.1953 | 2.99 | 150 | 0.1982 |
69
+ | 0.1953 | 3.19 | 160 | 0.1892 |
70
+ | 0.1953 | 3.39 | 170 | 0.1767 |
71
+ | 0.1953 | 3.59 | 180 | 0.1687 |
72
+ | 0.1953 | 3.78 | 190 | 0.1616 |
73
+ | 0.1469 | 3.98 | 200 | 0.1559 |
74
+ | 0.1469 | 4.18 | 210 | 0.1507 |
75
+ | 0.1469 | 4.38 | 220 | 0.1484 |
76
+ | 0.1469 | 4.58 | 230 | 0.1421 |
77
+ | 0.1469 | 4.78 | 240 | 0.1353 |
78
+ | 0.1212 | 4.98 | 250 | 0.1309 |
79
+ | 0.1212 | 5.18 | 260 | 0.1292 |
80
+ | 0.1212 | 5.38 | 270 | 0.1267 |
81
+ | 0.1212 | 5.58 | 280 | 0.1231 |
82
+ | 0.1212 | 5.78 | 290 | 0.1218 |
83
+ | 0.1059 | 5.98 | 300 | 0.1177 |
84
+ | 0.1059 | 6.18 | 310 | 0.1154 |
85
+ | 0.1059 | 6.37 | 320 | 0.1151 |
86
+ | 0.1059 | 6.57 | 330 | 0.1144 |
87
+ | 0.1059 | 6.77 | 340 | 0.1114 |
88
+ | 0.0936 | 6.97 | 350 | 0.1098 |
89
+ | 0.0936 | 7.17 | 360 | 0.1093 |
90
+ | 0.0936 | 7.37 | 370 | 0.1071 |
91
+ | 0.0936 | 7.57 | 380 | 0.1063 |
92
+ | 0.0936 | 7.77 | 390 | 0.1060 |
93
+ | 0.0881 | 7.97 | 400 | 0.1049 |
94
+ | 0.0881 | 8.17 | 410 | 0.1042 |
95
+ | 0.0881 | 8.37 | 420 | 0.1035 |
96
+ | 0.0881 | 8.57 | 430 | 0.1032 |
97
+ | 0.0881 | 8.76 | 440 | 0.1028 |
98
+ | 0.0819 | 8.96 | 450 | 0.1019 |
99
+ | 0.0819 | 9.16 | 460 | 0.1014 |
100
+ | 0.0819 | 9.36 | 470 | 0.1012 |
101
+ | 0.0819 | 9.56 | 480 | 0.1010 |
102
+ | 0.0819 | 9.76 | 490 | 0.1008 |
103
+ | 0.079 | 9.96 | 500 | 0.1008 |
104
 
105
 
106
  ### Framework versions
adapter_config.json CHANGED
@@ -21,10 +21,10 @@
21
  "target_modules": [
22
  "k_proj",
23
  "q_proj",
24
- "v_proj",
25
  "fc2",
26
- "dense",
27
- "fc1"
 
28
  ],
29
  "task_type": "CAUSAL_LM"
30
  }
 
21
  "target_modules": [
22
  "k_proj",
23
  "q_proj",
 
24
  "fc2",
25
+ "v_proj",
26
+ "fc1",
27
+ "dense"
28
  ],
29
  "task_type": "CAUSAL_LM"
30
  }
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:68d52b9ee5f1f1b896b6562f771ede7ebf1806ad405654c4f70b3cd896ce5c4f
3
  size 94422368
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0893ceadbbb8b438bd07e6464913f985f2fa6107ff3d249ce2e1a812c4cd4e1c
3
  size 94422368
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d4e6f08bbc3be4f8ef90b14fec37406472b33edd35f297e638593e1cbf0bbb83
3
  size 4283
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:30bdac43ad19d93ba5e682a664188d3572cb9c34d72d5639b1b67e5719607246
3
  size 4283