73 lines
2.1 KiB
YAML
73 lines
2.1 KiB
YAML
# ==========================================
|
|
# ACTUAL TEST CASES
|
|
# ==========================================
|
|
|
|
test_cases:
|
|
- name: "MiniMax-M2.5-w8a8"
|
|
model: "Eco-Tech/MiniMax-M2.5-w8a8-QuaRot"
|
|
envs:
|
|
HCCL_BUFFSIZE: "512"
|
|
HCCL_OP_EXPANSION_MODE: "AIV"
|
|
PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
|
|
OMP_NUM_THREADS: "1"
|
|
TASK_QUEUE_ENABLE: "1"
|
|
VLLM-ASCEND_ENABLE_NZ: "1"
|
|
VLLM_ASCEND_ENABLE_FLASHCOMM1: "1"
|
|
VLLM-ASCEND_BALANCE_SCHEDULING: "1"
|
|
VLLM_USE_MODELSCOPE: "true"
|
|
SERVER_PORT: "DEFAULT_PORT"
|
|
server_cmd:
|
|
- "--tensor-parallel-size"
|
|
- "8"
|
|
- "--data-parallel-size"
|
|
- "1"
|
|
- "--port"
|
|
- "$SERVER_PORT"
|
|
- "--trust-remote-code"
|
|
- "--gpu-memory-utilization"
|
|
- "0.85"
|
|
- "--quantization"
|
|
- "ascend"
|
|
- "--no-enable-prefix-caching"
|
|
- "--additional-config"
|
|
- '{"enable_cpu_binding":true}'
|
|
- "--speculative_config"
|
|
- '{"method":"eagle3","model":"vllm-ascend/MiniMax-M2.5-eagle-model-0318","num_speculative_tokens":3}'
|
|
- "--enable-expert-parallel"
|
|
- "--max-num-seqs"
|
|
- "128"
|
|
- "--max-num-batched-tokens"
|
|
- "16384"
|
|
- "--max-model-len"
|
|
- "196608"
|
|
- "--compilation-config"
|
|
- '{"cudagraph_mode":"FULL_DECODE_ONLY"}'
|
|
benchmarks:
|
|
acc_aime2025:
|
|
case_type: accuracy
|
|
dataset_path: vllm-ascend/aime2025
|
|
request_conf: vllm_api_general_chat
|
|
dataset_conf: aime2025/aime2025_gen_0_shot_chat_prompt
|
|
max_out_len: 131072
|
|
batch_size: 32
|
|
baseline: 90
|
|
threshold: 10
|
|
bos_token_id: 200019
|
|
do_sample: true
|
|
eos_token_id: 200020
|
|
temperature: 1.0
|
|
top_p: 0.95
|
|
top_k: 40
|
|
transformers_version: 4.46.1
|
|
ignore_eos: false
|
|
perf:
|
|
case_type: performance
|
|
dataset_path: vllm-ascend/GSM8K-in3500-bs2800
|
|
request_conf: vllm_api_stream_chat
|
|
dataset_conf: gsm8k/gsm8k_gen_0_shot_cot_str_perf
|
|
num_prompts: 512
|
|
max_out_len: 1500
|
|
batch_size: 128
|
|
request_rate: 0
|
|
baseline: 1116
|
|
threshold: 0.97 |