| # 2 * 80GB | |
| OMP_NUM_THREADS=14 \ | |
| CUDA_VISIBLE_DEVICES=0,1 \ | |
| swift export \ | |
| --model Qwen/Qwen2-57B-A14B-Instruct \ | |
| --dataset 'AI-ModelScope/alpaca-gpt4-data-zh#1000' \ | |
| 'AI-ModelScope/alpaca-gpt4-data-en#1000' \ | |
| --quant_n_samples 512 \ | |
| --quant_batch_size 1 \ | |
| --max_length 4096 \ | |
| --quant_method gptq \ | |
| --quant_bits 4 \ | |
| --output_dir Qwen2-57B-A14B-Instruct-GPTQ-Int4 | |