wehub-resource-sync
10 lines
242 B
Bash
10 lines
242 B
Bash
MODEL_PATH=$1
|
|
NUM_GPUS=4
|
|
BATCH_SIZE=60
|
|
PROMPT_TEMPLATE="alpaca"
|
|
|
|
python -m eval_vllm.driver \
|
|
--model_name_or_path $MODEL_PATH \
|
|
--batch_size $BATCH_SIZE \
|
|
--tensor_parallel_size $NUM_GPUS \
|
|
--prompt_template $PROMPT_TEMPLATE |