Hugging Face's logo

enkhtogtokh
/
mistral-dpo

TensorBoard
Safetensors
trl
dpo
Generated from Trainer
Model card Files Metrics Community
enkhtogtokh commited on
Commit
8f85bfb
·
verified ·
1 Parent(s): 79f1596

enkhtogtokh/mistral-dpom

Browse files
Files changed (5) hide show
  1. README.md +15 -35
  2. adapter_config.json +2 -2
  3. adapter_model.safetensors +1 -1
  4. runs/Jan15_14-36-59_d6f2f96c2d0f/events.out.tfevents.1705329535.d6f2f96c2d0f.7927.0 +3 -0
  5. training_args.bin +1 -1
README.md CHANGED
@@ -17,15 +17,15 @@ should probably proofread and complete it, then remove this comment. -->
17
 
18
  This model is a fine-tuned version of [TheBloke/OpenHermes-2-Mistral-7B-GPTQ](https://huggingface.co/TheBloke/OpenHermes-2-Mistral-7B-GPTQ) on the None dataset.
19
  It achieves the following results on the evaluation set:
20
- - Loss: 0.8386
21
- - Rewards/chosen: 1.3959
22
- - Rewards/rejected: 1.2944
23
- - Rewards/accuracies: 0.6635
24
- - Rewards/margins: 0.1015
25
- - Logps/rejected: -204.8275
26
- - Logps/chosen: -212.7901
27
- - Logits/rejected: -2.3737
28
- - Logits/chosen: -2.4043
29
 
30
  ## Model description
31
 
@@ -51,38 +51,18 @@ The following hyperparameters were used during training:
51
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
52
  - lr_scheduler_type: linear
53
  - lr_scheduler_warmup_steps: 2
54
- - training_steps: 250
55
  - mixed_precision_training: Native AMP
56
 
57
  ### Training results
58
 
59
  | Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
60
  |:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
61
- | 0.7034 | 0.0 | 10 | 0.6881 | 0.0669 | 0.0544 | 0.6442 | 0.0125 | -217.2274 | -226.0799 | -2.3083 | -2.3446 |
62
- | 0.6817 | 0.0 | 20 | 0.7081 | -0.0529 | -0.0428 | 0.5096 | -0.0100 | -218.1994 | -227.2778 | -2.3270 | -2.3588 |
63
- | 0.8729 | 0.0 | 30 | 0.7720 | -0.3001 | -0.2331 | 0.5 | -0.0670 | -220.1020 | -229.7500 | -2.3539 | -2.3796 |
64
- | 0.8805 | 0.0 | 40 | 0.7841 | -0.2908 | -0.2089 | 0.4712 | -0.0819 | -219.8597 | -229.6567 | -2.3597 | -2.3866 |
65
- | 0.7656 | 0.0 | 50 | 0.7334 | 0.0924 | 0.1060 | 0.5288 | -0.0136 | -216.7109 | -225.8250 | -2.3542 | -2.3835 |
66
- | 0.7447 | 0.0 | 60 | 0.7034 | 0.4187 | 0.3651 | 0.5288 | 0.0536 | -214.1196 | -222.5618 | -2.3546 | -2.3821 |
67
- | 0.5761 | 0.01 | 70 | 0.6868 | 0.7922 | 0.6818 | 0.6635 | 0.1105 | -210.9534 | -218.8266 | -2.3600 | -2.3880 |
68
- | 0.6459 | 0.01 | 80 | 0.6960 | 0.9805 | 0.8607 | 0.6154 | 0.1198 | -209.1636 | -216.9435 | -2.3727 | -2.4040 |
69
- | 0.6212 | 0.01 | 90 | 0.7216 | 1.0026 | 0.9052 | 0.5962 | 0.0974 | -208.7185 | -216.7225 | -2.3769 | -2.4083 |
70
- | 0.5771 | 0.01 | 100 | 0.7343 | 1.2182 | 1.1107 | 0.625 | 0.1075 | -206.6636 | -214.5667 | -2.3794 | -2.4081 |
71
- | 1.105 | 0.01 | 110 | 0.7612 | 1.3164 | 1.2262 | 0.625 | 0.0901 | -205.5086 | -213.5851 | -2.3803 | -2.4103 |
72
- | 0.5704 | 0.01 | 120 | 0.7712 | 1.2350 | 1.1516 | 0.6154 | 0.0834 | -206.2548 | -214.3988 | -2.3789 | -2.4092 |
73
- | 0.651 | 0.01 | 130 | 0.7840 | 1.2433 | 1.1676 | 0.6058 | 0.0757 | -206.0951 | -214.3164 | -2.3799 | -2.4109 |
74
- | 0.5913 | 0.01 | 140 | 0.7916 | 1.1579 | 1.0827 | 0.6058 | 0.0752 | -206.9440 | -215.1704 | -2.3800 | -2.4116 |
75
- | 0.7816 | 0.01 | 150 | 0.7879 | 1.0975 | 1.0127 | 0.6058 | 0.0849 | -207.6444 | -215.7737 | -2.3767 | -2.4085 |
76
- | 0.7859 | 0.01 | 160 | 0.7833 | 0.9808 | 0.8809 | 0.6346 | 0.0999 | -208.9623 | -216.9414 | -2.3730 | -2.4050 |
77
- | 0.6368 | 0.01 | 170 | 0.8020 | 1.1103 | 1.0068 | 0.6538 | 0.1035 | -207.7034 | -215.6462 | -2.3739 | -2.4052 |
78
- | 0.7555 | 0.01 | 180 | 0.8232 | 1.3013 | 1.2030 | 0.6731 | 0.0983 | -205.7407 | -213.7359 | -2.3795 | -2.4099 |
79
- | 1.0436 | 0.02 | 190 | 0.8346 | 1.3330 | 1.2393 | 0.6635 | 0.0938 | -205.3784 | -213.4188 | -2.3772 | -2.4080 |
80
- | 0.8714 | 0.02 | 200 | 0.8394 | 1.4131 | 1.3150 | 0.6731 | 0.0980 | -204.6207 | -212.6184 | -2.3790 | -2.4098 |
81
- | 0.5898 | 0.02 | 210 | 0.8321 | 1.4170 | 1.3099 | 0.6731 | 0.1072 | -204.6723 | -212.5786 | -2.3751 | -2.4056 |
82
- | 0.5896 | 0.02 | 220 | 0.8264 | 1.4029 | 1.2923 | 0.6827 | 0.1106 | -204.8480 | -212.7195 | -2.3727 | -2.4031 |
83
- | 0.7008 | 0.02 | 230 | 0.8307 | 1.4009 | 1.2911 | 0.6731 | 0.1098 | -204.8596 | -212.7401 | -2.3724 | -2.4028 |
84
- | 2.6272 | 0.02 | 240 | 0.8366 | 1.3934 | 1.2906 | 0.6635 | 0.1028 | -204.8651 | -212.8147 | -2.3735 | -2.4039 |
85
- | 0.9059 | 0.02 | 250 | 0.8386 | 1.3959 | 1.2944 | 0.6635 | 0.1015 | -204.8275 | -212.7901 | -2.3737 | -2.4043 |
86
 
87
 
88
  ### Framework versions
 
17
 
18
  This model is a fine-tuned version of [TheBloke/OpenHermes-2-Mistral-7B-GPTQ](https://huggingface.co/TheBloke/OpenHermes-2-Mistral-7B-GPTQ) on the None dataset.
19
  It achieves the following results on the evaluation set:
20
+ - Loss: 0.0002
21
+ - Rewards/chosen: 24.3722
22
+ - Rewards/rejected: -6.8482
23
+ - Rewards/accuracies: 1.0
24
+ - Rewards/margins: 31.2205
25
+ - Logps/rejected: -88.3652
26
+ - Logps/chosen: -328.0153
27
+ - Logits/rejected: -1.3660
28
+ - Logits/chosen: -1.8558
29
 
30
  ## Model description
31
 
 
51
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
52
  - lr_scheduler_type: linear
53
  - lr_scheduler_warmup_steps: 2
54
+ - training_steps: 50
55
  - mixed_precision_training: Native AMP
56
 
57
  ### Training results
58
 
59
  | Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
60
  |:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
61
+ | 0.5974 | 0.01 | 10 | 0.3220 | 2.6702 | -0.4084 | 0.9712 | 3.0786 | -23.9665 | -545.0353 | -1.2378 | -1.5696 |
62
+ | 0.0879 | 0.02 | 20 | 0.0217 | 13.3520 | -3.0537 | 1.0 | 16.4057 | -50.4196 | -438.2177 | -1.2934 | -1.7050 |
63
+ | 0.0413 | 0.03 | 30 | 0.0015 | 20.3280 | -5.4777 | 1.0 | 25.8057 | -74.6603 | -368.4581 | -1.3375 | -1.8145 |
64
+ | 0.0003 | 0.04 | 40 | 0.0003 | 23.8990 | -6.4549 | 1.0 | 30.3539 | -84.4315 | -332.7477 | -1.3562 | -1.8484 |
65
+ | 0.0002 | 0.05 | 50 | 0.0002 | 24.3722 | -6.8482 | 1.0 | 31.2205 | -88.3652 | -328.0153 | -1.3660 | -1.8558 |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
66
 
67
 
68
  ### Framework versions
adapter_config.json CHANGED
@@ -19,8 +19,8 @@
19
  "rank_pattern": {},
20
  "revision": null,
21
  "target_modules": [
22
- "q_proj",
23
- "v_proj"
24
  ],
25
  "task_type": "CAUSAL_LM"
26
  }
 
19
  "rank_pattern": {},
20
  "revision": null,
21
  "target_modules": [
22
+ "v_proj",
23
+ "q_proj"
24
  ],
25
  "task_type": "CAUSAL_LM"
26
  }
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:42b1b58c214146e1ef7e4fba8f7e8372f453587e7537ae7fe1283b9fe29751ba
3
  size 6832600
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2690627b1ee8d50ca1e9089ca5636f40a6038c42e2782441ecd4e1b1f0f6a681
3
  size 6832600
runs/Jan15_14-36-59_d6f2f96c2d0f/events.out.tfevents.1705329535.d6f2f96c2d0f.7927.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ebf061059a94c6c43782541b5fa5fc295eba0b524f4a97b4a990b434c3fc0157
3
+ size 12559
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:855d694fed9e062d72458809eb91cec0a8c065e1ae7665772773c22b3eb1c231
3
  size 4091
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dd825ceee038058b8a2f5bece56f2f93a6bcb66ea5a6992b366df62855309b21
3
  size 4091