dimasik1987 commited on
Commit
16130b9
·
verified ·
1 Parent(s): 719e010

End of training

Browse files
Files changed (2) hide show
  1. README.md +163 -0
  2. adapter_model.bin +3 -0
README.md ADDED
@@ -0,0 +1,163 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ base_model: NousResearch/Yarn-Llama-2-13b-64k
4
+ tags:
5
+ - axolotl
6
+ - generated_from_trainer
7
+ model-index:
8
+ - name: 8df67e46-c827-4414-a69b-f3a3598e33c3
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: NousResearch/Yarn-Llama-2-13b-64k
22
+ bf16: auto
23
+ chat_template: llama3
24
+ dataset_prepared_path: null
25
+ datasets:
26
+ - data_files:
27
+ - 70827c72e7e20cba_train_data.json
28
+ ds_type: json
29
+ format: custom
30
+ path: /workspace/input_data/70827c72e7e20cba_train_data.json
31
+ type:
32
+ field_input: ''
33
+ field_instruction: question
34
+ field_output: answer
35
+ format: '{instruction}'
36
+ no_input_format: '{instruction}'
37
+ system_format: '{system}'
38
+ system_prompt: ''
39
+ debug: null
40
+ deepspeed: null
41
+ device: cuda
42
+ early_stopping_patience: null
43
+ eval_max_new_tokens: 128
44
+ eval_table_size: null
45
+ evals_per_epoch: 4
46
+ flash_attention: false
47
+ fp16: null
48
+ fsdp: null
49
+ fsdp_config: null
50
+ gradient_accumulation_steps: 4
51
+ gradient_checkpointing: true
52
+ group_by_length: false
53
+ hub_model_id: dimasik1987/8df67e46-c827-4414-a69b-f3a3598e33c3
54
+ hub_repo: null
55
+ hub_strategy: checkpoint
56
+ hub_token: null
57
+ learning_rate: 0.0001
58
+ load_in_4bit: false
59
+ load_in_8bit: false
60
+ local_rank: null
61
+ logging_steps: 1
62
+ lora_alpha: 64
63
+ lora_dropout: 0.05
64
+ lora_fan_in_fan_out: null
65
+ lora_model_dir: null
66
+ lora_r: 32
67
+ lora_target_linear: true
68
+ lr_scheduler: cosine
69
+ max_memory:
70
+ 0: 70GiB
71
+ max_steps: 50
72
+ micro_batch_size: 2
73
+ mlflow_experiment_name: /tmp/70827c72e7e20cba_train_data.json
74
+ model_type: AutoModelForCausalLM
75
+ num_epochs: 3
76
+ optimizer: adamw_torch
77
+ output_dir: miner_id_24
78
+ pad_to_sequence_len: true
79
+ resume_from_checkpoint: null
80
+ s2_attention: null
81
+ sample_packing: false
82
+ save_steps: 25
83
+ save_strategy: steps
84
+ sequence_len: 2028
85
+ strict: false
86
+ tf32: false
87
+ tokenizer_type: AutoTokenizer
88
+ torch_dtype: bfloat16
89
+ train_on_inputs: false
90
+ trust_remote_code: true
91
+ val_set_size: 0.05
92
+ wandb_entity: null
93
+ wandb_mode: online
94
+ wandb_name: 8df67e46-c827-4414-a69b-f3a3598e33c3
95
+ wandb_project: Gradients-On-Demand
96
+ wandb_run: your_name
97
+ wandb_runid: 8df67e46-c827-4414-a69b-f3a3598e33c3
98
+ warmup_steps: 10
99
+ weight_decay: 0.01
100
+ xformers_attention: null
101
+
102
+ ```
103
+
104
+ </details><br>
105
+
106
+ # 8df67e46-c827-4414-a69b-f3a3598e33c3
107
+
108
+ This model is a fine-tuned version of [NousResearch/Yarn-Llama-2-13b-64k](https://huggingface.co/NousResearch/Yarn-Llama-2-13b-64k) on the None dataset.
109
+ It achieves the following results on the evaluation set:
110
+ - Loss: 0.1546
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: 0.0001
130
+ - train_batch_size: 2
131
+ - eval_batch_size: 2
132
+ - seed: 42
133
+ - gradient_accumulation_steps: 4
134
+ - total_train_batch_size: 8
135
+ - optimizer: Use OptimizerNames.ADAMW_TORCH 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: 50
139
+
140
+ ### Training results
141
+
142
+ | Training Loss | Epoch | Step | Validation Loss |
143
+ |:-------------:|:------:|:----:|:---------------:|
144
+ | 20.7654 | 0.0008 | 1 | 4.9974 |
145
+ | 18.0608 | 0.0042 | 5 | 4.3991 |
146
+ | 5.6643 | 0.0084 | 10 | 0.8882 |
147
+ | 0.8245 | 0.0126 | 15 | 0.1505 |
148
+ | 0.0053 | 0.0168 | 20 | 0.1627 |
149
+ | 0.0024 | 0.0211 | 25 | 0.1644 |
150
+ | 1.8089 | 0.0253 | 30 | 0.1860 |
151
+ | 0.3383 | 0.0295 | 35 | 0.1672 |
152
+ | 0.0192 | 0.0337 | 40 | 0.1700 |
153
+ | 0.8968 | 0.0379 | 45 | 0.1611 |
154
+ | 0.337 | 0.0421 | 50 | 0.1546 |
155
+
156
+
157
+ ### Framework versions
158
+
159
+ - PEFT 0.13.2
160
+ - Transformers 4.46.0
161
+ - Pytorch 2.5.0+cu124
162
+ - Datasets 3.0.1
163
+ - Tokenizers 0.20.1
adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5d3198412e9d6a9ffcb9d8072144f3cd945707a41bc593c05f32eed123dec667
3
+ size 500897546