Hugging Face's logo

enkhtogtokh
/
mistral-dpo

TensorBoard
Safetensors
trl
dpo
Generated from Trainer
Model card Files Metrics Community
enkhtogtokh commited on
Commit
79f1596
·
verified ·
1 Parent(s): 5165845

enkhtogtokh/mistral-dpo

Browse files
Files changed (10) hide show
  1. README.md +93 -0
  2. adapter_config.json +26 -0
  3. adapter_model.safetensors +3 -0
  4. added_tokens.json +4 -0
  5. runs/Jan15_08-21-38_2c69626c07f9/events.out.tfevents.1705307174.2c69626c07f9.144.0 +3 -0
  6. special_tokens_map.json +35 -0
  7. tokenizer.json +0 -0
  8. tokenizer.model +3 -0
  9. tokenizer_config.json +64 -0
  10. training_args.bin +3 -0
README.md ADDED
@@ -0,0 +1,93 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ base_model: TheBloke/OpenHermes-2-Mistral-7B-GPTQ
4
+ tags:
5
+ - trl
6
+ - dpo
7
+ - generated_from_trainer
8
+ model-index:
9
+ - name: mistral-dpo
10
+ results: []
11
+ ---
12
+
13
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
14
+ should probably proofread and complete it, then remove this comment. -->
15
+
16
+ # mistral-dpo
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
+
32
+ More information needed
33
+
34
+ ## Intended uses & limitations
35
+
36
+ More information needed
37
+
38
+ ## Training and evaluation data
39
+
40
+ More information needed
41
+
42
+ ## Training procedure
43
+
44
+ ### Training hyperparameters
45
+
46
+ The following hyperparameters were used during training:
47
+ - learning_rate: 0.0002
48
+ - train_batch_size: 1
49
+ - eval_batch_size: 8
50
+ - seed: 42
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
89
+
90
+ - Transformers 4.35.2
91
+ - Pytorch 2.0.1+cu117
92
+ - Datasets 2.15.0
93
+ - Tokenizers 0.15.0
adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "TheBloke/OpenHermes-2-Mistral-7B-GPTQ",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": true,
9
+ "layers_pattern": null,
10
+ "layers_to_transform": null,
11
+ "loftq_config": {},
12
+ "lora_alpha": 6,
13
+ "lora_dropout": 0.1,
14
+ "megatron_config": null,
15
+ "megatron_core": "megatron.core",
16
+ "modules_to_save": null,
17
+ "peft_type": "LORA",
18
+ "r": 4,
19
+ "rank_pattern": {},
20
+ "revision": null,
21
+ "target_modules": [
22
+ "q_proj",
23
+ "v_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:42b1b58c214146e1ef7e4fba8f7e8372f453587e7537ae7fe1283b9fe29751ba
3
+ size 6832600
added_tokens.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "<|im_end|>": 32000,
3
+ "<|im_start|>": 32001
4
+ }
runs/Jan15_08-21-38_2c69626c07f9/events.out.tfevents.1705307174.2c69626c07f9.144.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:321b47960c64c3b7a625d1f95eca78247dd898fccfd02f0e55e0a124bb2036e0
3
+ size 39878
special_tokens_map.json ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<unk>",
4
+ "<s>",
5
+ "</s>"
6
+ ],
7
+ "bos_token": {
8
+ "content": "<s>",
9
+ "lstrip": false,
10
+ "normalized": false,
11
+ "rstrip": false,
12
+ "single_word": false
13
+ },
14
+ "eos_token": {
15
+ "content": "<|im_end|>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false
20
+ },
21
+ "pad_token": {
22
+ "content": "</s>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false
27
+ },
28
+ "unk_token": {
29
+ "content": "<unk>",
30
+ "lstrip": false,
31
+ "normalized": false,
32
+ "rstrip": false,
33
+ "single_word": false
34
+ }
35
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055
3
+ size 493443
tokenizer_config.json ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "added_tokens_decoder": {
5
+ "0": {
6
+ "content": "<unk>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "1": {
14
+ "content": "<s>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "2": {
22
+ "content": "</s>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ },
29
+ "32000": {
30
+ "content": "<|im_end|>",
31
+ "lstrip": false,
32
+ "normalized": false,
33
+ "rstrip": false,
34
+ "single_word": false,
35
+ "special": true
36
+ },
37
+ "32001": {
38
+ "content": "<|im_start|>",
39
+ "lstrip": false,
40
+ "normalized": false,
41
+ "rstrip": false,
42
+ "single_word": false,
43
+ "special": false
44
+ }
45
+ },
46
+ "additional_special_tokens": [
47
+ "<unk>",
48
+ "<s>",
49
+ "</s>"
50
+ ],
51
+ "bos_token": "<s>",
52
+ "clean_up_tokenization_spaces": false,
53
+ "eos_token": "<|im_end|>",
54
+ "legacy": true,
55
+ "model_max_length": 1000000000000000019884624838656,
56
+ "pad_token": "</s>",
57
+ "sp_model_kwargs": {},
58
+ "spaces_between_special_tokens": false,
59
+ "tokenizer_class": "LlamaTokenizer",
60
+ "trust_remote_code": false,
61
+ "unk_token": "<unk>",
62
+ "use_default_system_prompt": true,
63
+ "use_fast": true
64
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:855d694fed9e062d72458809eb91cec0a8c065e1ae7665772773c22b3eb1c231
3
+ size 4091