Hugging Face's logo

RAIJAY
/
NBERT_DPO

PEFT
Safetensors
trl
dpo
Generated from Trainer
Model card Files Community
RAIJAY commited on
Commit
29abc24
•
1 Parent(s): 301a47c

RAIJAY/NBERT_2

Browse files
Files changed (4) hide show
  1. README.md +21 -16
  2. adapter_config.json +5 -5
  3. adapter_model.safetensors +1 -1
  4. training_args.bin +1 -1
README.md CHANGED
@@ -18,15 +18,15 @@ should probably proofread and complete it, then remove this comment. -->
18
 
19
  This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on the None dataset.
20
  It achieves the following results on the evaluation set:
21
- - Loss: 0.0000
22
- - Rewards/chosen: 7.1665
23
- - Rewards/rejected: -23.6077
24
- - Rewards/accuracies: 1.0
25
- - Rewards/margins: 30.7742
26
- - Logps/rejected: -457.5194
27
- - Logps/chosen: -360.8431
28
- - Logits/rejected: -3.9255
29
- - Logits/chosen: -3.9174
30
 
31
  ## Model description
32
 
@@ -54,16 +54,21 @@ The following hyperparameters were used during training:
54
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
55
  - lr_scheduler_type: linear
56
  - lr_scheduler_warmup_steps: 2
57
- - training_steps: 1600
58
 
59
  ### Training results
60
 
61
- | Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
62
- |:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
63
- | 0.0308 | 1.0 | 400 | 0.0528 | -26.5381 | -56.0808 | 0.9950 | 29.5427 | -782.2505 | -697.8889 | -1.9518 | -2.1901 |
64
- | 0.0056 | 2.0 | 800 | 0.0000 | 6.9898 | -23.7179 | 1.0 | 30.7077 | -458.6215 | -362.6105 | -3.9255 | -3.9191 |
65
- | 0.0 | 3.0 | 1200 | 0.0000 | 7.1599 | -23.6095 | 1.0 | 30.7694 | -457.5371 | -360.9089 | -3.9257 | -3.9175 |
66
- | 0.0 | 4.0 | 1600 | 0.0000 | 7.1665 | -23.6077 | 1.0 | 30.7742 | -457.5194 | -360.8431 | -3.9255 | -3.9174 |
 
 
 
 
 
67
 
68
 
69
  ### Framework versions
 
18
 
19
  This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on the None dataset.
20
  It achieves the following results on the evaluation set:
21
+ - Loss: 0.0229
22
+ - Rewards/chosen: -740.3076
23
+ - Rewards/rejected: -1059.6395
24
+ - Rewards/accuracies: 0.9988
25
+ - Rewards/margins: 319.3320
26
+ - Logps/rejected: -10817.7158
27
+ - Logps/chosen: -7838.3896
28
+ - Logits/rejected: -32.6170
29
+ - Logits/chosen: -26.3151
30
 
31
  ## Model description
32
 
 
54
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
55
  - lr_scheduler_type: linear
56
  - lr_scheduler_warmup_steps: 2
57
+ - training_steps: 25806
58
 
59
  ### Training results
60
 
61
+ | Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
62
+ |:-------------:|:-----:|:-----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
63
+ | 37.7926 | 0.67 | 2867 | 2.8916 | -149.7567 | -197.0506 | 0.9359 | 47.2939 | -2191.8269 | -1932.8804 | 38.0669 | 4.1036 |
64
+ | 84.7229 | 1.33 | 5734 | 2.0247 | -327.4202 | -450.9706 | 0.9656 | 123.5504 | -4731.0264 | -3709.5146 | -17.6232 | -16.2614 |
65
+ | 0.4302 | 2.0 | 8601 | 0.2490 | -391.4300 | -536.8747 | 0.9923 | 145.4447 | -5590.0679 | -4349.6123 | -13.6537 | -12.7337 |
66
+ | 0.6952 | 2.67 | 11468 | 0.0587 | -606.4489 | -775.4740 | 0.9970 | 169.0251 | -7976.0605 | -6499.8027 | 8.1646 | -0.2018 |
67
+ | 0.2119 | 3.33 | 14335 | 0.2843 | -641.6364 | -925.0908 | 0.9907 | 283.4543 | -9472.2285 | -6851.6772 | -11.2088 | -13.0496 |
68
+ | 0.129 | 4.0 | 17202 | 0.1065 | -706.7910 | -1019.4420 | 0.9958 | 312.6511 | -10415.7412 | -7503.2227 | 29.4650 | 10.0032 |
69
+ | 0.1046 | 4.67 | 20069 | 0.1005 | -758.2514 | -1105.3041 | 0.9977 | 347.0525 | -11274.3594 | -8017.8281 | -37.3526 | -28.3912 |
70
+ | 0.0656 | 5.33 | 22936 | 0.0241 | -790.2775 | -1078.3324 | 0.9986 | 288.0548 | -11004.6445 | -8338.0889 | -7.1017 | -13.6854 |
71
+ | 0.0 | 6.0 | 25803 | 0.0229 | -740.3076 | -1059.6395 | 0.9988 | 319.3320 | -10817.7158 | -7838.3896 | -32.6170 | -26.3151 |
72
 
73
 
74
  ### Framework versions
adapter_config.json CHANGED
@@ -19,14 +19,14 @@
19
  "rank_pattern": {},
20
  "revision": null,
21
  "target_modules": [
22
- "k_proj",
23
- "v_proj",
24
- "o_proj",
25
  "down_proj",
26
- "q_proj",
27
  "lm_head",
28
  "up_proj",
29
- "gate_proj"
 
 
 
30
  ],
31
  "task_type": "CAUSAL_LM"
32
  }
 
19
  "rank_pattern": {},
20
  "revision": null,
21
  "target_modules": [
 
 
 
22
  "down_proj",
23
+ "gate_proj",
24
  "lm_head",
25
  "up_proj",
26
+ "q_proj",
27
+ "v_proj",
28
+ "o_proj",
29
+ "k_proj"
30
  ],
31
  "task_type": "CAUSAL_LM"
32
  }
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:93b06d7bec2e223edc14b5f5cc1ccf52509afb9e7edb3c9c9d4e1e898e91f152
3
  size 609389712
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ae0f871bb072a5dd8c81e42087e6de9788c7501cd0d3f965e5c58e8158f835cd
3
  size 609389712
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:fac80a1406018f1cb0005fa5e9b342c56a623e3a800c66bc0510d263d4f5e39f
3
  size 4219
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:068c33107d83952fcc7d63d650e0d78843ebe13a25b623183abfa4c8a25c5396
3
  size 4219