Hugging Face's logo

romainnn
/
2da3d576-a2e0-43ca-ab56-ba6262085721

PEFT
Safetensors
mistral
axolotl
Generated from Trainer
Model card Files
xet
 Community
romainnn commited on
Commit
54e2910
·
verified ·
1 Parent(s): 3ebb60c

End of training

Browse files
Files changed (3) hide show
  1. README.md +161 -0
  2. adapter_model.bin +3 -0
  3. adapter_model.safetensors +1 -1
README.md ADDED
@@ -0,0 +1,161 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ base_model: jhflow/mistral7b-lora-multi-turn-v2
4
+ tags:
5
+ - axolotl
6
+ - generated_from_trainer
7
+ model-index:
8
+ - name: 2da3d576-a2e0-43ca-ab56-ba6262085721
9
+ results: []
10
+ ---
11
+
12
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
13
+ should probably proofread and complete it, then remove this comment. -->
14
+
15
+ [<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)
16
+ <details><summary>See axolotl config</summary>
17
+
18
+ axolotl version: `0.4.1`
19
+ ```yaml
20
+ adapter: lora
21
+ base_model: jhflow/mistral7b-lora-multi-turn-v2
22
+ bf16: true
23
+ chat_template: llama3
24
+ dataset_prepared_path: null
25
+ datasets:
26
+ - data_files:
27
+ - e59bf08a82c586dc_train_data.json
28
+ ds_type: json
29
+ format: custom
30
+ path: /workspace/input_data/e59bf08a82c586dc_train_data.json
31
+ type:
32
+ field_instruction: ja
33
+ field_output: en
34
+ format: '{instruction}'
35
+ no_input_format: '{instruction}'
36
+ system_format: '{system}'
37
+ system_prompt: ''
38
+ debug: null
39
+ deepspeed: null
40
+ early_stopping_patience: 2
41
+ eval_max_new_tokens: 128
42
+ eval_steps: 100
43
+ eval_table_size: null
44
+ flash_attention: true
45
+ fp16: null
46
+ fsdp: null
47
+ fsdp_config: null
48
+ gradient_accumulation_steps: 8
49
+ gradient_checkpointing: true
50
+ group_by_length: false
51
+ hub_model_id: romainnn/2da3d576-a2e0-43ca-ab56-ba6262085721
52
+ hub_repo: null
53
+ hub_strategy: checkpoint
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
+ lr_scheduler: cosine
72
+ max_grad_norm: 1.0
73
+ max_steps: 840
74
+ micro_batch_size: 4
75
+ mlflow_experiment_name: /tmp/e59bf08a82c586dc_train_data.json
76
+ model_type: AutoModelForCausalLM
77
+ num_epochs: 2
78
+ optimizer: adamw_bnb_8bit
79
+ output_dir: miner_id_24
80
+ pad_to_sequence_len: true
81
+ resume_from_checkpoint: null
82
+ s2_attention: null
83
+ sample_packing: false
84
+ save_steps: 100
85
+ sequence_len: 2048
86
+ strict: false
87
+ tf32: true
88
+ tokenizer_type: AutoTokenizer
89
+ train_on_inputs: false
90
+ trust_remote_code: true
91
+ val_set_size: 0.03384644544629923
92
+ wandb_entity: null
93
+ wandb_mode: online
94
+ wandb_name: 495a38a9-bfac-4582-90e8-fd041e7f79aa
95
+ wandb_project: Gradients-On-Demand
96
+ wandb_run: your_name
97
+ wandb_runid: 495a38a9-bfac-4582-90e8-fd041e7f79aa
98
+ warmup_steps: 10
99
+ weight_decay: 0.0
100
+ xformers_attention: null
101
+
102
+ ```
103
+
104
+ </details><br>
105
+
106
+ # 2da3d576-a2e0-43ca-ab56-ba6262085721
107
+
108
+ This model is a fine-tuned version of [jhflow/mistral7b-lora-multi-turn-v2](https://huggingface.co/jhflow/mistral7b-lora-multi-turn-v2) on the None dataset.
109
+ It achieves the following results on the evaluation set:
110
+ - Loss: 0.6719
111
+
112
+ ## Model description
113
+
114
+ More information needed
115
+
116
+ ## Intended uses & limitations
117
+
118
+ More information needed
119
+
120
+ ## Training and evaluation data
121
+
122
+ More information needed
123
+
124
+ ## Training procedure
125
+
126
+ ### Training hyperparameters
127
+
128
+ The following hyperparameters were used during training:
129
+ - learning_rate: 0.0002
130
+ - train_batch_size: 4
131
+ - eval_batch_size: 4
132
+ - seed: 42
133
+ - gradient_accumulation_steps: 8
134
+ - total_train_batch_size: 32
135
+ - optimizer: Use OptimizerNames.ADAMW_BNB with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
136
+ - lr_scheduler_type: cosine
137
+ - lr_scheduler_warmup_steps: 10
138
+ - training_steps: 840
139
+
140
+ ### Training results
141
+
142
+ | Training Loss | Epoch | Step | Validation Loss |
143
+ |:-------------:|:------:|:----:|:---------------:|
144
+ | 20.535 | 0.0002 | 1 | 2.3938 |
145
+ | 7.7195 | 0.0224 | 100 | 0.8167 |
146
+ | 5.0283 | 0.0448 | 200 | 0.7787 |
147
+ | 5.4587 | 0.0673 | 300 | 0.7641 |
148
+ | 6.4127 | 0.0897 | 400 | 0.7289 |
149
+ | 6.5381 | 0.1121 | 500 | 0.7130 |
150
+ | 5.3058 | 0.1345 | 600 | 0.6915 |
151
+ | 4.8455 | 0.1569 | 700 | 0.6764 |
152
+ | 6.4455 | 0.1794 | 800 | 0.6719 |
153
+
154
+
155
+ ### Framework versions
156
+
157
+ - PEFT 0.13.2
158
+ - Transformers 4.46.0
159
+ - Pytorch 2.5.0+cu124
160
+ - Datasets 3.0.1
161
+ - Tokenizers 0.20.1
adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9eba72d8f33f1080e7d0c79d7b9a4d04bae5cba5f9225a6d194eec904d4f2e80
3
+ size 167934026
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:596f4aa8cb31fba7bb294f4253c9196aa48a56a4fe7e9745f0bfe798dd583748
3
  size 167832240
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:67f9a2c87d6aa73b88374cb917b8413a776dc3988ad245348d96b3f16abe64e3
3
  size 167832240