12 lines
260 B
Bash
12 lines
260 B
Bash
CUDA_VISIBLE_DEVICES=0 \
|
|
swift export \
|
|
--model Qwen/Qwen3-30B-A3B \
|
|
--quant_method fp8 \
|
|
--output_dir Qwen3-30B-A3B-FP8
|
|
|
|
# CUDA_VISIBLE_DEVICES=0 \
|
|
# swift infer \
|
|
# --model Qwen3-30B-A3B-FP8 \
|
|
# --infer_backend vllm \
|
|
# --stream true
|