leixa commited on
Commit
8023149
·
verified ·
1 Parent(s): 447fee8

End of training

Browse files
Files changed (2) hide show
  1. README.md +161 -0
  2. adapter_model.bin +3 -0
README.md ADDED
@@ -0,0 +1,161 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ license: other
4
+ base_model: Qwen/Qwen1.5-0.5B-Chat
5
+ tags:
6
+ - axolotl
7
+ - generated_from_trainer
8
+ model-index:
9
+ - name: 85fddd35-46cd-43ac-9700-55395a4f0703
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: Qwen/Qwen1.5-0.5B-Chat
23
+ bf16: auto
24
+ chat_template: llama3
25
+ dataset_prepared_path: null
26
+ datasets:
27
+ - data_files:
28
+ - b38b9d3a1551c71d_train_data.json
29
+ ds_type: json
30
+ format: custom
31
+ path: /workspace/input_data/b38b9d3a1551c71d_train_data.json
32
+ type:
33
+ field_instruction: pt
34
+ field_output: vmw
35
+ format: '{instruction}'
36
+ no_input_format: '{instruction}'
37
+ system_format: '{system}'
38
+ system_prompt: ''
39
+ debug: null
40
+ deepspeed: null
41
+ early_stopping_patience: null
42
+ eval_max_new_tokens: 128
43
+ eval_table_size: null
44
+ evals_per_epoch: 4
45
+ flash_attention: false
46
+ fp16: true
47
+ fsdp: null
48
+ fsdp_config: null
49
+ gradient_accumulation_steps: 4
50
+ gradient_checkpointing: true
51
+ group_by_length: false
52
+ hub_model_id: leixa/85fddd35-46cd-43ac-9700-55395a4f0703
53
+ hub_repo: null
54
+ hub_strategy: checkpoint
55
+ hub_token: null
56
+ learning_rate: 0.0001
57
+ load_in_4bit: false
58
+ load_in_8bit: false
59
+ local_rank: 0
60
+ logging_steps: 3
61
+ lora_alpha: 128
62
+ lora_dropout: 0.1
63
+ lora_fan_in_fan_out: true
64
+ lora_model_dir: null
65
+ lora_r: 64
66
+ lora_target_linear: true
67
+ lr_scheduler: cosine
68
+ max_memory:
69
+ 0: 72GB
70
+ max_steps: 100
71
+ micro_batch_size: 8
72
+ mlflow_experiment_name: /tmp/b38b9d3a1551c71d_train_data.json
73
+ model_type: AutoModelForCausalLM
74
+ num_epochs: 3
75
+ optimizer: adamw_bnb_8bit
76
+ output_dir: miner_id_24
77
+ pad_to_sequence_len: true
78
+ resume_from_checkpoint: null
79
+ s2_attention: false
80
+ sample_packing: false
81
+ saves_per_epoch: 4
82
+ sequence_len: 1024
83
+ strict: false
84
+ tf32: false
85
+ tokenizer_type: AutoTokenizer
86
+ train_on_inputs: false
87
+ trust_remote_code: true
88
+ val_set_size: 0.05
89
+ wandb_entity: leixa-personal
90
+ wandb_mode: online
91
+ wandb_name: 85fddd35-46cd-43ac-9700-55395a4f0703
92
+ wandb_project: Gradients-On-Demand
93
+ wandb_run: your_name
94
+ wandb_runid: 85fddd35-46cd-43ac-9700-55395a4f0703
95
+ warmup_steps: 10
96
+ weight_decay: 0.01
97
+ xformers_attention: null
98
+
99
+ ```
100
+
101
+ </details><br>
102
+
103
+ # 85fddd35-46cd-43ac-9700-55395a4f0703
104
+
105
+ This model is a fine-tuned version of [Qwen/Qwen1.5-0.5B-Chat](https://huggingface.co/Qwen/Qwen1.5-0.5B-Chat) on the None dataset.
106
+ It achieves the following results on the evaluation set:
107
+ - Loss: 3.7726
108
+
109
+ ## Model description
110
+
111
+ More information needed
112
+
113
+ ## Intended uses & limitations
114
+
115
+ More information needed
116
+
117
+ ## Training and evaluation data
118
+
119
+ More information needed
120
+
121
+ ## Training procedure
122
+
123
+ ### Training hyperparameters
124
+
125
+ The following hyperparameters were used during training:
126
+ - learning_rate: 0.0001
127
+ - train_batch_size: 8
128
+ - eval_batch_size: 8
129
+ - seed: 42
130
+ - gradient_accumulation_steps: 4
131
+ - total_train_batch_size: 32
132
+ - optimizer: Use OptimizerNames.ADAMW_BNB with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
133
+ - lr_scheduler_type: cosine
134
+ - lr_scheduler_warmup_steps: 10
135
+ - training_steps: 100
136
+
137
+ ### Training results
138
+
139
+ | Training Loss | Epoch | Step | Validation Loss |
140
+ |:-------------:|:------:|:----:|:---------------:|
141
+ | No log | 0.0186 | 1 | 7.1221 |
142
+ | 6.2763 | 0.1674 | 9 | 5.9018 |
143
+ | 5.2322 | 0.3349 | 18 | 5.0585 |
144
+ | 4.6 | 0.5023 | 27 | 4.6171 |
145
+ | 4.3017 | 0.6698 | 36 | 4.3168 |
146
+ | 4.0419 | 0.8372 | 45 | 4.1112 |
147
+ | 4.5154 | 1.0093 | 54 | 3.9870 |
148
+ | 3.8323 | 1.1767 | 63 | 3.9053 |
149
+ | 3.5854 | 1.3442 | 72 | 3.8268 |
150
+ | 3.8804 | 1.5116 | 81 | 3.7939 |
151
+ | 3.6054 | 1.6791 | 90 | 3.7757 |
152
+ | 3.5648 | 1.8465 | 99 | 3.7726 |
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:a26ad48d131a1b2c8dd8d78e03d2aa04503a87607a52056bfa0559eabc6aa7c8
3
+ size 121232010