Views
No views yet
0.4.11base_model: Crystalcareai/Meta-llama-3.1-8b-instruct
2bf16: 'True'
3chat_template: chatml
4dataset_prepared_path: ./sft_processed
5dataset_processes: 12
6datasets:
7- field_messages: messages
8 path: sft_dataset
9 type: sharegpt
10deepspeed: /axolotl/deepspeed_configs/zero3_bf16.json
11eval_batch_size: 1
12flash_attention: true
13gradient_accumulation_steps: 8
14gradient_checkpointing: true
15learning_rate: 5.0e-06
16logging_steps: 1
17lr_scheduler: cosine
18micro_batch_size: 2
19num_epochs: 3
20optimizer: adamw_bnb_8bit
21output_dir: data/sft-full
22pad_to_sequence_len: true
23sample_packing: true
24save_safetensors: true
25save_total_limit: 0
26saves_per_epoch: 0
27seed: 42
28sequence_len: 4096
29special_tokens:
30 pad_token: <|end_of_text|>
31tf32: false
32tokens: []
33use_tensorboard: true
34val_set_size: 0
35