246 lines
9.8 KiB
Python
246 lines
9.8 KiB
Python
"""
|
||
xc_validation_strategy — 主入口
|
||
|
||
启动后执行一次模型验证任务批量提交，之后保持 HTTP 服务存活。
|
||
同时暴露 /health（K8s 探活）和 /status（运行状态）。
|
||
"""
|
||
|
||
import json
|
||
import os
|
||
import signal
|
||
import threading
|
||
from datetime import datetime
|
||
from http.server import BaseHTTPRequestHandler, ThreadingHTTPServer
|
||
from typing import List, Tuple
|
||
|
||
import requests
|
||
|
||
# ══════════════════════════════════════════════════════════
|
||
# 配置（全部从环境变量读取，不硬编码敏感信息）
|
||
# ══════════════════════════════════════════════════════════
|
||
BASE_URL = os.environ.get("BASE_URL", "https://modelhub.org.cn")
|
||
SUBMIT_ENDPOINT = "/adminApi/async/task/create-contest-task"
|
||
|
||
# 通过 curl -X POST https://modelhub.org.cn/adminApi/user/login 获取后填入
|
||
AUTH_TOKEN = "eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJ1c2VyQWNjb3VudCI6Inpob3VzaGFzaGEiLCJpZCI6MTQsInVzZXJSb2xlIjoibGVhZGVyYm9hcmQiLCJleHAiOjE3ODE4NTE0NzcsImlhdCI6MTc4MTI0NjY3N30.p3uvCpG50aLNifNVVXxvzmWJahbLM5K1671FVCtj8E8"
|
||
CONTEST_API_TOKEN = "ef1ef82f3c9efee413d602345fbe224d"
|
||
CONTRIBUTORS = "zhoushasha"
|
||
GPU_TYPE = "Cambricon_mlu-370-x8"
|
||
TASK_TYPE = "text-generation"
|
||
STRATEGY_ID = os.environ.get("STRATEGY_ID", "") # 平台自动注入，无需修改
|
||
|
||
HTTP_HOST = "0.0.0.0"
|
||
HTTP_PORT = 8080
|
||
|
||
# ══════════════════════════════════════════════════════════
|
||
# 模型列表
|
||
# ══════════════════════════════════════════════════════════
|
||
ALL_MODEL_IDS = [
|
||
"l3utterfly/mistral-7b-v0.1-layla-v4",
|
||
"OpenBuddy/openbuddy-mistral-7b-v13.1",
|
||
"allenai/truthfulqa-info-judge-llama2-7B",
|
||
"l3utterfly/mistral-7b-v0.1-layla-v1",
|
||
"l3utterfly/minima-3b-layla-v2",
|
||
"l3utterfly/tinyllama-1.1b-layla-v4",
|
||
"l3utterfly/mistral-7b-v0.1-layla-v2",
|
||
"l3utterfly/tinyllama-1.1b-layla-v1",
|
||
"Duxiaoman-DI/XuanYuan-13B-Chat",
|
||
"l3utterfly/minima-3b-layla-v1",
|
||
"AI-ModelScope/gemma-2-2b",
|
||
"baichuan-inc/Baichuan-13B-Base",
|
||
"LGAI-EXAONE/EXAONE-Deep-2.4B",
|
||
"NousResearch/DeepHermes-3-Llama-3-3B-Preview",
|
||
"Fengshenbang/Ziya2-13B-Base",
|
||
"prithivMLmods/QwQ-MathOct-7B",
|
||
"l3utterfly/phi-2-layla-v1-chatml",
|
||
"argilla/notus-7b-v1",
|
||
"prithivMLmods/Doopler-Augment-3B-Cox",
|
||
"prithivMLmods/Blaze.1-32B-Instruct",
|
||
"CohereLabs/aya-expanse-8B",
|
||
"Magpie-Align/MagpieLM-4B-SFT-v0.1",
|
||
"Magpie-Align/MagpieLM-8B-SFT-v0.1",
|
||
"Magpie-Align/Llama-3-8B-Magpie-Align-SFT-v0.2",
|
||
"Magpie-Align/MagpieLM-8B-Chat-v0.1",
|
||
"Magpie-Align/Llama-3.1-8B-Magpie-Align-SFT-v0.1",
|
||
"Magpie-Align/Llama-3-8B-Magpie-Air-SFT-300K-v0.1",
|
||
"prithivMLmods/Tulu-MathLingo-8B",
|
||
"prithivMLmods/Triangulum-5B",
|
||
"prithivMLmods/Viper-Coder-v0.1",
|
||
]
|
||
|
||
# ══════════════════════════════════════════════════════════
|
||
# 全局状态（供 /status 展示）
|
||
# ══════════════════════════════════════════════════════════
|
||
_state = {
|
||
"strategy_id": STRATEGY_ID,
|
||
"phase": "starting", # starting | submitting | done | error
|
||
"total": len(ALL_MODEL_IDS),
|
||
"submitted": 0,
|
||
"failed": 0,
|
||
"started_at": None,
|
||
"finished_at": None,
|
||
}
|
||
_shutdown = threading.Event()
|
||
|
||
# ══════════════════════════════════════════════════════════
|
||
# HTTP 服务
|
||
# ══════════════════════════════════════════════════════════
|
||
class Handler(BaseHTTPRequestHandler):
|
||
def do_GET(self):
|
||
if self.path == "/health":
|
||
self._json({"status": "ok"})
|
||
elif self.path == "/status":
|
||
self._json(_state)
|
||
else:
|
||
self._json({"error": "not found"}, 404)
|
||
|
||
def _json(self, body: dict, code: int = 200):
|
||
payload = json.dumps(body, default=str).encode()
|
||
self.send_response(code)
|
||
self.send_header("Content-Type", "application/json")
|
||
self.send_header("Content-Length", str(len(payload)))
|
||
self.end_headers()
|
||
self.wfile.write(payload)
|
||
|
||
def log_message(self, fmt, *args):
|
||
print(f"[http] {self.address_string()} {fmt % args}", flush=True)
|
||
|
||
|
||
def _run_http():
|
||
server = ThreadingHTTPServer((HTTP_HOST, HTTP_PORT), Handler)
|
||
server.timeout = 1
|
||
print(f"[http] 监听 {HTTP_HOST}:{HTTP_PORT}", flush=True)
|
||
while not _shutdown.is_set():
|
||
server.handle_request()
|
||
server.server_close()
|
||
print("[http] 已关闭", flush=True)
|
||
|
||
# ══════════════════════════════════════════════════════════
|
||
# 业务逻辑
|
||
# ══════════════════════════════════════════════════════════
|
||
def _submit_task(token: str, model_id: str) -> Tuple[bool, str]:
|
||
headers = {
|
||
"Content-Type": "application/json",
|
||
"Authorization": f"Bearer {token}",
|
||
}
|
||
config_content = f"""docker_image: harbor.4pd.io/hardcore-tech/cambricon-mlu370-pytorch:v25.01-torch2.5.0-torchmlu1.24.1-ubuntu22.04-py310
|
||
nv_docker_image: harbor.4pd.io/dooke/vllm/vllm/vllm-openai:v0.11.0
|
||
framework: vllm
|
||
storage: gpfs
|
||
modelhub_options:
|
||
srcRelativePath: leaderboard/modelHubXC/{model_id}
|
||
mountPoint: /model
|
||
sut_config:
|
||
values:
|
||
gpu_num: 1
|
||
env:
|
||
- name: MAX_MODEL_LEN
|
||
value: 8192
|
||
command: ["vllm", "serve", "/model", "--port", "8000", "--served-model-name", "llm", "--max-model-len", "8192", "--trust-remote-code", "--dtype", "float16"]
|
||
ref_config:
|
||
values:
|
||
cpu_num: 2
|
||
gpu_num: 1
|
||
env:
|
||
- name: MAX_MODEL_LEN
|
||
value: 8192
|
||
command: ["vllm", "serve", "/model", "--port", "80", "--served-model-name", "llm", "--max-model-len", "8192", "--trust-remote-code", "--dtype", "float16"]
|
||
"""
|
||
payload = {
|
||
"contestApiToken": CONTEST_API_TOKEN,
|
||
"contributors": CONTRIBUTORS,
|
||
"gpuTypes": [GPU_TYPE],
|
||
"taskType": TASK_TYPE,
|
||
"modelId": model_id,
|
||
"framework": "vllm",
|
||
"strategyId": STRATEGY_ID, # 平台要求
|
||
"submissionConfig": [{
|
||
"config": config_content,
|
||
"gpuType": GPU_TYPE,
|
||
"taskType": TASK_TYPE,
|
||
}],
|
||
}
|
||
print(f"[payload] {json.dumps(payload, indent=2, ensure_ascii=False)}", flush=True)
|
||
try:
|
||
resp = requests.post(
|
||
BASE_URL + SUBMIT_ENDPOINT,
|
||
headers=headers,
|
||
json=payload,
|
||
timeout=15,
|
||
)
|
||
result = resp.json()
|
||
if result.get("code") == 0:
|
||
task_id = result.get("data", {}).get("id", "")
|
||
print(f"[worker] OK {model_id} task_id={task_id}", flush=True)
|
||
return True, task_id
|
||
else:
|
||
print(f"[worker] FAIL {model_id}: {result.get('message')}", flush=True)
|
||
return False, ""
|
||
except Exception as e:
|
||
print(f"[worker] ERROR {model_id}: {e}", flush=True)
|
||
return False, ""
|
||
|
||
|
||
def _run_worker():
|
||
_state["started_at"] = datetime.utcnow().isoformat()
|
||
_state["phase"] = "submitting"
|
||
|
||
successful: List[Tuple[str, str]] = []
|
||
token = AUTH_TOKEN
|
||
print("[worker] 使用预设 Token，跳过登录", flush=True)
|
||
|
||
for model_id in ALL_MODEL_IDS:
|
||
if _shutdown.is_set():
|
||
break
|
||
ok, task_id = _submit_task(token, model_id)
|
||
if ok:
|
||
_state["submitted"] += 1
|
||
successful.append((task_id, model_id))
|
||
else:
|
||
_state["failed"] += 1
|
||
|
||
# 写入结果文件
|
||
try:
|
||
with open("submitted_validation_tasks.txt", "w", encoding="utf-8") as f:
|
||
for tid, mid in successful:
|
||
f.write(f"{tid}\t{mid}\n")
|
||
except Exception:
|
||
pass
|
||
|
||
_state["finished_at"] = datetime.utcnow().isoformat()
|
||
_state["phase"] = "done"
|
||
print(
|
||
f"[worker] 完成 submitted={_state['submitted']} failed={_state['failed']}",
|
||
flush=True,
|
||
)
|
||
# 提交完成后继续保持进程存活，等待平台停止
|
||
|
||
# ══════════════════════════════════════════════════════════
|
||
# 入口
|
||
# ══════════════════════════════════════════════════════════
|
||
def _handle_signal(signum, _frame):
|
||
print(f"[main] 收到信号 {signum}，正在关闭...", flush=True)
|
||
_shutdown.set()
|
||
|
||
|
||
def main():
|
||
signal.signal(signal.SIGTERM, _handle_signal)
|
||
signal.signal(signal.SIGINT, _handle_signal)
|
||
|
||
# HTTP 服务线程
|
||
http_thread = threading.Thread(target=_run_http, daemon=False)
|
||
http_thread.start()
|
||
|
||
# 提交任务线程
|
||
worker_thread = threading.Thread(target=_run_worker, daemon=True)
|
||
worker_thread.start()
|
||
|
||
# 主线程等待 shutdown
|
||
_shutdown.wait()
|
||
print("[main] 等待 HTTP 服务关闭...", flush=True)
|
||
http_thread.join(timeout=5)
|
||
print("[main] 退出", flush=True)
|
||
|
||
|
||
if __name__ == "__main__":
|
||
main() |