wehub-resource-sync
15 lines
415 B
Bash
15 lines
415 B
Bash
pip install "transformers<4.52"
|
|
|
|
CUDA_VISIBLE_DEVICES=0 \
|
|
swift export \
|
|
--model Qwen/Qwen2.5-72B-Instruct \
|
|
--dataset 'AI-ModelScope/alpaca-gpt4-data-zh#500' \
|
|
'AI-ModelScope/alpaca-gpt4-data-en#500' \
|
|
--device_map cpu \
|
|
--quant_n_samples 256 \
|
|
--quant_batch_size 1 \
|
|
--max_length 2048 \
|
|
--quant_method awq \
|
|
--quant_bits 4 \
|
|
--output_dir Qwen2.5-72B-Instruct-AWQ
|