Hugging Face's logo

Alphatao
/
7ea479b6-461c-4338-8f81-b7982e6f4067

PEFT
Safetensors
mistral
axolotl
Generated from Trainer
Model card Files
xet
 Community
Alphatao commited on
Commit
a2e34c0
·
verified ·
1 Parent(s): 1425440

End of training

Browse files
Files changed (3) hide show
  1. README.md +164 -0
  2. adapter_model.bin +3 -0
  3. adapter_model.safetensors +1 -1
README.md ADDED
@@ -0,0 +1,164 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ license: apache-2.0
4
+ base_model: teknium/OpenHermes-2.5-Mistral-7B
5
+ tags:
6
+ - axolotl
7
+ - generated_from_trainer
8
+ model-index:
9
+ - name: 7ea479b6-461c-4338-8f81-b7982e6f4067
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
+ [<img src="https://raw.githubusercontent.com/axolotl-ai-cloud/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/axolotl-ai-cloud/axolotl)
17
+ <details><summary>See axolotl config</summary>
18
+
19
+ axolotl version: `0.4.1`
20
+ ```yaml
21
+ adapter: lora
22
+ base_model: teknium/OpenHermes-2.5-Mistral-7B
23
+ bf16: true
24
+ chat_template: llama3
25
+ dataset_prepared_path: null
26
+ datasets:
27
+ - data_files:
28
+ - 82f6a0053cefa75a_train_data.json
29
+ ds_type: json
30
+ format: custom
31
+ path: /workspace/input_data/82f6a0053cefa75a_train_data.json
32
+ type:
33
+ field_instruction: input
34
+ field_output: reference_answer
35
+ format: '{instruction}'
36
+ no_input_format: '{instruction}'
37
+ system_format: '{system}'
38
+ system_prompt: ''
39
+ debug: null
40
+ device_map:
41
+ ? ''
42
+ : 0,1,2,3,4,5,6,7
43
+ early_stopping_patience: 2
44
+ eval_max_new_tokens: 128
45
+ eval_steps: 100
46
+ eval_table_size: null
47
+ flash_attention: true
48
+ gradient_accumulation_steps: 8
49
+ gradient_checkpointing: true
50
+ group_by_length: false
51
+ hub_model_id: Alphatao/7ea479b6-461c-4338-8f81-b7982e6f4067
52
+ hub_repo: null
53
+ hub_strategy: null
54
+ hub_token: null
55
+ learning_rate: 0.0002
56
+ load_best_model_at_end: true
57
+ load_in_4bit: false
58
+ load_in_8bit: false
59
+ local_rank: null
60
+ logging_steps: 1
61
+ lora_alpha: 32
62
+ lora_dropout: 0.05
63
+ lora_fan_in_fan_out: null
64
+ lora_model_dir: null
65
+ lora_r: 16
66
+ lora_target_linear: true
67
+ lora_target_modules:
68
+ - q_proj
69
+ - k_proj
70
+ - v_proj
71
+ - o_proj
72
+ - down_proj
73
+ - up_proj
74
+ lr_scheduler: cosine
75
+ max_grad_norm: 1.0
76
+ max_steps: 658
77
+ micro_batch_size: 4
78
+ mlflow_experiment_name: /tmp/82f6a0053cefa75a_train_data.json
79
+ model_type: AutoModelForCausalLM
80
+ num_epochs: 2
81
+ optimizer: adamw_bnb_8bit
82
+ output_dir: miner_id_24
83
+ pad_to_sequence_len: true
84
+ resume_from_checkpoint: null
85
+ s2_attention: null
86
+ sample_packing: false
87
+ save_steps: 100
88
+ sequence_len: 2048
89
+ special_tokens:
90
+ pad_token: <|im_end|>
91
+ strict: false
92
+ tf32: true
93
+ tokenizer_type: AutoTokenizer
94
+ train_on_inputs: false
95
+ trust_remote_code: true
96
+ val_set_size: 0.05
97
+ wandb_entity: null
98
+ wandb_mode: online
99
+ wandb_name: 99fe5060-7d88-43a9-8f08-86e6aceedc15
100
+ wandb_project: Gradients-On-Demand
101
+ wandb_run: your_name
102
+ wandb_runid: 99fe5060-7d88-43a9-8f08-86e6aceedc15
103
+ warmup_steps: 10
104
+ weight_decay: 0.0
105
+ xformers_attention: null
106
+
107
+ ```
108
+
109
+ </details><br>
110
+
111
+ # 7ea479b6-461c-4338-8f81-b7982e6f4067
112
+
113
+ This model is a fine-tuned version of [teknium/OpenHermes-2.5-Mistral-7B](https://huggingface.co/teknium/OpenHermes-2.5-Mistral-7B) on the None dataset.
114
+ It achieves the following results on the evaluation set:
115
+ - Loss: 0.0006
116
+
117
+ ## Model description
118
+
119
+ More information needed
120
+
121
+ ## Intended uses & limitations
122
+
123
+ More information needed
124
+
125
+ ## Training and evaluation data
126
+
127
+ More information needed
128
+
129
+ ## Training procedure
130
+
131
+ ### Training hyperparameters
132
+
133
+ The following hyperparameters were used during training:
134
+ - learning_rate: 0.0002
135
+ - train_batch_size: 4
136
+ - eval_batch_size: 4
137
+ - seed: 42
138
+ - gradient_accumulation_steps: 8
139
+ - total_train_batch_size: 32
140
+ - optimizer: Use OptimizerNames.ADAMW_BNB with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
141
+ - lr_scheduler_type: cosine
142
+ - lr_scheduler_warmup_steps: 10
143
+ - training_steps: 658
144
+
145
+ ### Training results
146
+
147
+ | Training Loss | Epoch | Step | Validation Loss |
148
+ |:-------------:|:------:|:----:|:---------------:|
149
+ | 10.1359 | 0.0004 | 1 | 1.2132 |
150
+ | 1.5158 | 0.0445 | 100 | 0.1932 |
151
+ | 0.4393 | 0.0889 | 200 | 0.0454 |
152
+ | 0.0508 | 0.1334 | 300 | 0.0144 |
153
+ | 0.046 | 0.1779 | 400 | 0.0062 |
154
+ | 0.0025 | 0.2223 | 500 | 0.0014 |
155
+ | 0.004 | 0.2668 | 600 | 0.0006 |
156
+
157
+
158
+ ### Framework versions
159
+
160
+ - PEFT 0.13.2
161
+ - Transformers 4.46.0
162
+ - Pytorch 2.5.0+cu124
163
+ - Datasets 3.0.1
164
+ - Tokenizers 0.20.1
adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a5e469c451f20480dcbb9a277aaa213e8eb63d449f08857a3101643d44aea5b8
3
+ size 167934026
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:03e5f6012507ab64ff7b8e42090fb1d76473323181bc77cacbfa3028c3664440
3
  size 167832240
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:31ed0033fbffb9a30399f905d76f82a882227f3fda59ab57799d9509df882ba1
3
  size 167832240