llama3_gptq.yaml 311 Bytes
Newer Older
chenych's avatar
chenych committed
1
2
3
### model
model_name_or_path: meta-llama/Meta-Llama-3-8B-Instruct
template: llama3
luopl's avatar
luopl committed
4
trust_remote_code: true
chenych's avatar
chenych committed
5
6

### export
chenych's avatar
chenych committed
7
export_dir: output/llama3_gptq
chenych's avatar
chenych committed
8
export_quantization_bit: 4
chenych's avatar
chenych committed
9
export_quantization_dataset: data/c4_demo.jsonl
chenych's avatar
chenych committed
10
export_size: 5
chenych's avatar
chenych committed
11
export_device: cpu  # choices: [cpu, auto]
chenych's avatar
chenych committed
12
export_legacy_format: false