fb8e22ec00
### What this PR does / why we need it?
1. Add nightly test on MiniMax-M2.5 with deployment method on A3
2. Add MiniMax-M2.5 deployment introduction to vllm-ascend docs
- vLLM version: v0.17.0
- vLLM main:
4034c3d32e
---------
Signed-off-by: limuyuan <limuyuan3@huawei.com>
Signed-off-by: SparrowMu <52023119+SparrowMu@users.noreply.github.com>
Co-authored-by: limuyuan <limuyuan3@huawei.com>
47 lines
1.3 KiB
YAML
47 lines
1.3 KiB
YAML
# ==========================================
|
|
# ACTUAL TEST CASES
|
|
# ==========================================
|
|
|
|
test_cases:
|
|
- name: "MiniMax-M2.5-TP16-Reasoning-Tool"
|
|
model: "MiniMax/MiniMax-M2.5"
|
|
envs:
|
|
HCCL_BUFFSIZE: "1024"
|
|
OMP_PROC_BIND: "false"
|
|
HCCL_OP_EXPANSION_MODE: "AIV"
|
|
PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
|
|
VLLM_ASCEND_ENABLE_FLASHCOMM1: "1"
|
|
SERVER_PORT: "DEFAULT_PORT"
|
|
prompts:
|
|
- "Hello. Please introduce yourself briefly."
|
|
api_keyword_args:
|
|
max_tokens: 128
|
|
temperature: 0
|
|
test_content:
|
|
- chat_completion
|
|
server_cmd:
|
|
- "--tensor-parallel-size"
|
|
- "16"
|
|
- "--port"
|
|
- "$SERVER_PORT"
|
|
- "--trust-remote-code"
|
|
- "--dtype"
|
|
- "bfloat16"
|
|
- "--enable-expert-parallel"
|
|
- "--max-num-seqs"
|
|
- "32"
|
|
- "--max-num-batched-tokens"
|
|
- "32768"
|
|
# Prefer a smaller max length for nightly stability. For full context,
|
|
# omit this flag and rely on the model config (196608).
|
|
- "--max-model-len"
|
|
- "32768"
|
|
- "--compilation-config"
|
|
- '{"cudagraph_mode":"FULL_DECODE_ONLY"}'
|
|
- "--enable-auto-tool-choice"
|
|
- "--tool-call-parser"
|
|
- "minimax_m2"
|
|
- "--reasoning-parser"
|
|
- "minimax_m2_append_think"
|
|
|