wehub-resource-sync
14 lines
408 B
Bash
14 lines
408 B
Bash
# 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
|