submit 338 nonquantized candidates on PPU and Hygon

Candidate pool is the 24711 already-downloaded models, which satisfies
mechanism A's precondition that the model be present in platform storage.

Strict and relaxed condition-2 give identical results this round: of the 24007
pooled models with verify records, 24006 already have 已验证 on some card, so
relaxing to include 验证中 adds just one. The binding constraint is now how few
models lack a record on each card - Kunlunxin_p-800, Cambricon_mlu-370-x8 and
Iluvatar_bi-150 are all at zero and are left out of GPU_JOBS.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

.gitignore

@@ -0,0 +1,2 @@
+.DS_Store
+__pycache__/

Dockerfile

@@ -1,11 +1,7 @@
 FROM modelhubxc-4pd.tencentcloudcr.com/xc_agent_platform/python:3.11-slim
 
-ENV PYTHONUNBUFFERED=1 \
+ENV PYTHONUNBUFFERED=1 
-    USER_ACCOUNT="zhoushasha@4paradigm.com" \
+    
-    USER_PASSWORD="4pdpassword" \
-    CONTEST_API_TOKEN="ef1ef82f3c9efee413d602345fbe224d" \
-    CONTRIBUTORS="zhoushasha" \
-    GPU_TYPE="Cambricon_mlu-370-x8"
 
 WORKDIR /app
 

README.md

@@ -1,5 +1,35 @@
 # xc_validation_strategy
 
-信创自动化模型适配平台 — 验证策略服务
+批量向 ModelHub XC 平台提交模型验证任务的策略服务，之后保持 HTTP 服务存活供平台探活。
 
-从 HuggingFace 周期性抓取新模型，自动完成同步、下载、提交验证任务的全流程，常驻运行在 xc_agent_platform 上。
+当前批次使用 2026-09-24 的成功下载模型清单，经非量化过滤后，要求目标 GPU 无验证记录、至少一张其他 GPU 的状态为「已验证」。仅提交 `ppu_zw_810e` 的 17 个模型和 `hygon_k100-ai` 的 321 个模型，共 338 个任务；上一批模型不会重复提交。
+
+## 功能
+
+- 使用 `AUTH_TOKEN` 环境变量（未提供时使用 `main.py` 中的预设 Bearer Token）向 `zhoushasha` 账号提交任务
+- 分 GPU 读取 `model_ids/` 中的模型清单，批量提交 vLLM 验证任务
+- 账号额度暂满时，每 30 分钟重试尚未提交的模型
+- 提交结果写入 `submitted_validation_tasks.txt`
+- 暴露 `/health` 和 `/status` 接口满足平台运行时契约
+
+## 项目结构
+
+```
+.
+├── main.py            # 主入口：HTTP 服务 + 提交逻辑
+├── model_ids/         # 本批次两张 GPU 的模型 ID 清单
+├── Dockerfile         # 平台镜像构建配置
+├── requirements.txt   # Python 依赖
+└── submitted_validation_tasks.txt   # 运行后自动生成，记录提交结果
+```
+
+## 平台契约说明
+
+本项目满足平台对策略镜像的全部必要约束：
+
+- Dockerfile 位于仓库根目录，基于官方轻量基础镜像
+- 暴露 8080 端口并实现 `GET /health`
+- 通过环境变量 `STRATEGY_ID` 获取策略 ID
+- 正确处理 `SIGTERM` 信号，支持优雅停机
+
+预设 Bearer Token 于 2026-10-01 22:13（北京时间）到期；届时如需再次运行，应更新 `AUTH_TOKEN`。

main.py

@@ -1,17 +1,28 @@
 """
 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。
+启动后针对 GPU_JOBS 中配置的 GPU 卡分别批量提交各自筛选出的模型验证任务。
-同时暴露 /health（K8s 探活）和 /status（运行状态）。
+本轮仅提交 ppu_zw_810e 和 hygon_k100-ai 的 2026-09-24 非量化候选清单；
+其他 GPU 的 config_content 模板仍保留，但未列入 GPU_JOBS。
+（/adminApi/async/task/create-contest-task，
+Bearer Token 认证），之后保持 HTTP 服务存活。
+
+账号额度自动重试：如果某个模型提交时命中"当前等待中或运行中的异步模型验证
+任务数量已达上限"（账号额度已满），不算永久失败，会被留到下一轮；额度耗尽后
+本进程会原地等待 30 分钟，再自动重试所有因额度问题未提交成功的模型，如此循环，
+直至全部提交成功或进程被平台关闭——不需要重新部署新策略，循环逻辑在本进程内完成。
+非额度原因的失败（如模型已在验证中等）不会重试。
+
+同时暴露 /health（K8s 探活）和 /status（运行状态，含当前轮次/待重试数/下次重试时间）。
 """
 
 import json
 import os
 import signal
 import threading
-import traceback
 from datetime import datetime
 from http.server import BaseHTTPRequestHandler, ThreadingHTTPServer
+from pathlib import Path
 from typing import List, Tuple
 
 import requests
@@ -20,57 +31,76 @@ 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")
-LOGIN_ENDPOINT     = "/adminApi/user/login"
 SUBMIT_ENDPOINT    = "/adminApi/async/task/create-contest-task"
 
-USER_ACCOUNT       = os.environ["USER_ACCOUNT"]        # 必填
+# 通过 curl -X POST https://modelhub.org.cn/adminApi/user/login 获取后填入
-USER_PASSWORD      = os.environ["USER_PASSWORD"]        # 必填
+AUTH_TOKEN        = os.environ.get("AUTH_TOKEN", "eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJ1c2VyQWNjb3VudCI6Inpob3VzaGFzaGEiLCJpZCI6MTQsInVzZXJSb2xlIjoibGVhZGVyYm9hcmQiLCJleHAiOjE3OTA4NjQwMDMsImlhdCI6MTc5MDI1OTIwM30.T23Tp3xcI8kkKIOwRCXmZlpe3Qo3sOIxZ8n6NbzJJ2M")
-CONTEST_API_TOKEN  = os.environ["CONTEST_API_TOKEN"]    # 必填
+CONTEST_API_TOKEN = "ef1ef82f3c9efee413d602345fbe224d"
-STRATEGY_ID        = os.environ.get("STRATEGY_ID", "")  # 平台注入
+CONTRIBUTORS      = "zhoushasha"
-CONTRIBUTORS       = os.environ.get("CONTRIBUTORS", USER_ACCOUNT)
+TASK_TYPE         = "text-generation"
-GPU_TYPE           = os.environ.get("GPU_TYPE", "Cambricon_mlu-370-x8")
+STRATEGY_ID       = os.environ.get("STRATEGY_ID", "")  # 平台自动注入，无需修改
-TASK_TYPE          = os.environ.get("TASK_TYPE", "text-generation")
 
 HTTP_HOST = "0.0.0.0"
 HTTP_PORT = 8080
 
 # ══════════════════════════════════════════════════════════
-#  模型列表
+#  各 GPU 的模型列表
 # ══════════════════════════════════════════════════════════
-ALL_MODEL_IDS = [
+MODEL_DIR = Path(__file__).resolve().parent / "model_ids"
-    "AI-ModelScope/gemma-2b",
+
-    "AI-ModelScope/falcon-mamba-7b",
+
-    "katanemo/deepseek-2",
+def load_model_ids(filename: str) -> List[str]:
-    "OpenBMB/MiniCPM4-0.5B",
+    model_ids = [
-    "NousResearch/Meta-Llama-3-8B-Instruct",
+        line.strip()
-    "MediaTek-Research/Breeze-7B-Instruct-v1_0",
+        for line in (MODEL_DIR / filename).read_text(encoding="utf-8").splitlines()
-    "QLUNLP/BianCang-Qwen2.5-7B-Instruct",
+        if line.strip()
-    "OpenBMB/MiniCPM4-Survey",
+    ]
-    "OpenBMB/MiniCPM4-8B",
+    if not model_ids or len(model_ids) != len(set(model_ids)):
-    "PaddlePaddle/ERNIE-4.5-0.3B-PT",
+        raise ValueError(f"模型列表为空或含重复 ID: {filename}")
-    "LLM-Research/Llama-Guard-3-8B",
+    return model_ids
-    "OpenBMB/MiniCPM-2B-dpo-fp16",
+
-    "OpenBMB/MiniCPM4.1-8B",
+
-    "Cylingo/Xinyuan-LLM-14B-0428",
+BIREN_MODELS = [
-    "Fengshenbang/Ziya-LLaMA-13B-v1",
+    "zipaltrivedi/dotnet-coder-14b",
-    "baichuan-inc/Baichuan2-13B-Chat",
-    "LLM-Research/gemma-2-9b-it",
-    "Qwen/CodeQwen1.5-7B-Chat",
-    "OpenBMB/cpm-bee-10b",
-    "OpenBMB/MiniCPM3-4B",
 ]
 
+CAMBRICON_MODELS = [
+]
+
+METAX_MODELS = [
+    "zipaltrivedi/dotnet-coder-14b",
+]
+
+HYGON_MODELS = load_model_ids("hygon_k100-ai_2026-09-24.txt")
+
+KUNLUNXIN_MODELS = [
+]
+
+PPU_MODELS = load_model_ids("ppu_zw_810e_2026-09-24.txt")
+
+# 本轮提交：2026-09-24 已成功下载、非量化、目标卡无验证记录，且其他卡至少一张「已验证」的模型。
+# 仅 ppu_zw_810e(17) 和 hygon_k100-ai(321)，共 338 个；不重复提交上一轮的 24 个任务。
+GPU_JOBS: List[Tuple[str, List[str]]] = [
+    ("ppu_zw_810e",          PPU_MODELS),
+    ("hygon_k100-ai",        HYGON_MODELS),
+]
+TOTAL_MODELS = sum(len(models) for _, models in GPU_JOBS)
+
 # ══════════════════════════════════════════════════════════
 #  全局状态（供 /status 展示）
 # ══════════════════════════════════════════════════════════
 _state = {
     "strategy_id":   STRATEGY_ID,
-    "phase":         "starting",   # starting | submitting | done | error
+    "phase":         "starting",   # starting | submitting | waiting_retry | done | error
-    "total":         len(ALL_MODEL_IDS),
+    "total":         TOTAL_MODELS,
     "submitted":     0,
     "failed":        0,
+    "per_gpu":       {gpu: 0 for gpu, _ in GPU_JOBS},
     "started_at":    None,
     "finished_at":   None,
+    "round":               0,     # 当前是第几轮提交
+    "quota_blocked_remaining": 0, # 因额度上限暂未提交成功、等待下一轮重试的模型数
+    "next_retry_at":       None,  # 下一轮重试的预计时间（额度耗尽等待期间）
 }
 _shutdown = threading.Event()
 
@@ -108,29 +138,40 @@ def _run_http():
     print("[http] 已关闭", flush=True)
 
 # ══════════════════════════════════════════════════════════
-#  业务逻辑
+#  各 GPU 的 config_content 模板
 # ══════════════════════════════════════════════════════════
-def _login() -> str:
+def build_config_content(gpu_type: str, model_id: str) -> str:
-    headers = {"Content-Type": "application/json"}
+    if gpu_type == "Biren_166m":
-    resp = requests.post(
+        max_model_len = 4096
-        BASE_URL + LOGIN_ENDPOINT,
+        return f"""docker_image: git.modelhub.org.cn:9443/enginex/xc-llm-biren166m:26.01
-        headers=headers,
+nv_docker_image: harbor.4pd.io/dooke/vllm/vllm/vllm-openai:v0.11.0
-        json={"userAccount": USER_ACCOUNT, "userPassword": USER_PASSWORD},
+framework: vllm
-        timeout=30,
+lang: zh
-    )
+storage: gpfs
-    data = resp.json()
+api: completion
-    if data.get("code") != 0:
+modelhub_options:
-        raise RuntimeError(f"登录失败: {data.get('message')}")
+  srcRelativePath: leaderboard/modelHubXC/{model_id}
-    print("[worker] 登录成功", flush=True)
+  mountPoint: /model
-    return data["data"]["token"]
+max_model_len: {max_model_len}
-
+sut_config:
-
+  values:
-def _submit_task(token: str, model_id: str) -> Tuple[bool, str]:
+    gpu_num: 1
-    headers = {
+    env:
-        "Content-Type": "application/json",
+      - name: MAX_MODEL_LEN
-        "Authorization": f"Bearer {token}",
+        value: {max_model_len}
-    }
+    command: ['/bin/bash', '-ic', 'vllm serve /model --port 8000 --served-model-name llm --max-model-len {max_model_len} --gpu-memory-utilization 0.9 --enforce-eager --trust-remote-code -tp 1 --host 0.0.0.0']
-    config_content = f"""docker_image: harbor.4pd.io/hardcore-tech/cambricon-mlu370-pytorch:v25.01-torch2.5.0-torchmlu1.24.1-ubuntu22.04-py310
+ref_config:
+  values:
+    cpu_num: 2
+    gpu_num: 1
+    env:
+      - name: MAX_MODEL_LEN
+        value: {max_model_len}
+    command: ['vllm', 'serve', '/model', '--port', '80', '--served-model-name', 'llm', '--max-model-len', '{max_model_len}', '--enforce-eager', '--trust-remote-code', '-tp', '1']
+model: llm
+"""
+    elif gpu_type == "Cambricon_mlu-370-x8":
+        return 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
@@ -153,20 +194,178 @@ ref_config:
         value: 8192
     command: ["vllm", "serve", "/model", "--port", "80", "--served-model-name", "llm", "--max-model-len", "8192", "--trust-remote-code", "--dtype", "float16"]
 """
+    elif gpu_type == "MetaX_c-500":
+        return f"""docker_image: git.modelhub.org.cn:9443/enginex-metax/vllm:0.9.1
+nv_docker_image: harbor.4pd.io/dooke/vllm/vllm/vllm-openai:v0.11.0
+framework: vllm
+lang: en
+storage: gpfs
+api: chat
+modelhub_options:
+  srcRelativePath: leaderboard/modelHubXC/{model_id}
+  mountPoint: /model
+max_model_len: 2048
+sut_config:
+  gpu_num: 1
+  values:
+    command: ['/opt/conda/bin/vllm', 'serve', '/model', '--port', '20644', '--served-model-name', 'llm',  '--max-model-len', '2048', '--gpu-memory-utilization', '0.9',  '--enforce-eager', '--trust-remote-code' ,'-tp', '1']
+ref_config:
+  gpu_num: 1
+  values:
+    command: ['vllm', 'serve', '/model', '--port', '80', '--served-model-name', 'llm', '--max-model-len', '2048', '--enforce-eager', '--trust-remote-code', '-tp', '1']
+"""
+    elif gpu_type == "Kunlunxin_p-800":
+        return f"""docker_image: git.modelhub.org.cn:9443/enginex/xc-llm-kunlun
+nv_docker_image: harbor.4pd.io/dooke/vllm/vllm/vllm-openai:v0.11.0
+framework: vllm
+lang: en
+storage: gpfs
+api: chat
+temperature: 0.4
+repetition_penalty: 1.1
+top_p: 0.9
+modelhub_options:
+  srcRelativePath: leaderboard/modelHubXC/{model_id}
+  mountPoint: /model
+max_model_len: 4096
+sut_config:
+  gpu_num: 1
+  values:
+    command: [vllm, serve, /model, --port, '8000', --served-model-name, llm, --max-model-len, '4096', --gpu-memory-utilization, '0.9', --enforce-eager, --trust-remote-code, -tp, '1']
+ref_config:
+  gpu_num: 1
+  values:
+    command: [vllm, serve, /model, --port, '80', --served-model-name, llm, --max-model-len, '4096', --enforce-eager, --trust-remote-code, -tp, '1']
+"""
+    elif gpu_type == "hygon_k100-ai":
+        return f"""
+docker_image: harbor.4pd.io/modelhubxc/enginex-hygon/vllm:0.9.2-patch-tokenizer
+nv_docker_image: harbor.4pd.io/modelhubxc/enginex-nvidia/vllm:0.11.0-patch-tokenizer
+framework: vllm
+storage: gpfs
+
+max_model_len: 4096
+sut_config:
+    gpu_num: 1
+    values:
+        command: ['vllm', 'serve', '/model', '--port', '20644', '--served-model-name', 'llm',  '--max-model-len', '4096', '--enforce-eager', '--trust-remote-code' ,'-tp', '1' ]
+ref_config:
+    gpu_num: 1
+    values:
+        command: ['vllm', 'serve', '/model', '--port', '80', '--served-model-name', 'llm', '--max-model-len', '4096', '--enforce-eager', '--trust-remote-code', '-tp', '1']
+"""
+    elif gpu_type == "ppu_zw_810e":
+        return f"""gpu_type: ppu_zw_810e
+framework: vllm
+docker_image: harbor.4pd.io/hardcore-tech/asllm:1.10.1-pytorch2.10.0-ubuntu24.04-sail2.1.0-cuda13.0-sglang0.5.10-vllm0.19.0-py312
+nv_docker_image: harbor-contest.4pd.io/sunruoxi/vllm-openai-fix-tokenizer:v0.11.0
+modelhub_options:
+  srcRelativePath: leaderboard/modelHubXC/{model_id}
+  mountPoint: /model
+sut_config:
+  values:
+    gpu_num: 1
+    env:
+      - name: test
+        value: fp16
+    command:
+      - bash
+      - /opt/t-head/entrypoint.sh
+      - python3
+      - -m
+      - asllm.entrypoints.api_server
+      - --model
+      - /model
+      - --port
+      - '30000'
+      - --host
+      - 0.0.0.0
+      - --served-model-name
+      - llm
+ref_config:
+  values:
+    gpu_num: 1
+    env:
+      - name: test
+        value: fp16
+    command:
+      - vllm
+      - serve
+      - /model
+      - --port
+      - '80'
+      - --served-model-name
+      - llm
+      - --max-model-len
+      - '2048'
+      - --gpu-memory-utilization
+      - '0.9'
+      - --enforce-eager
+      - --trust-remote-code
+      - -tp
+      - '1'
+"""
+    
+    elif gpu_type == "Iluvatar_bi-150":
+            return f"""docker_image: harbor-contest.4pd.io/luopingyi/enginex-iluvatar-bi150/vllm:0.8.3
+nv_docker_image: harbor.4pd.io/dooke/vllm/vllm/vllm-openai:v0.11.0
+framework: vllm
+api: completion
+temperature: 0.7
+repetition_penalty: 1.2
+top_p: 0.9
+
+max_model_len: 4096
+max_tokens: 1024
+sut_config:
+  gpu_num: 1
+  values:
+    command: ['vllm', 'serve', '/model', '--port', '80', '--served-model-name', 'llm', '--max-model-len', '4096', '--gpu-memory-utilization', '0.9', '--enforce-eager', '--trust-remote-code' ,'-tp', '1']
+ref_config:
+  gpu_num: 1
+  values:
+    command: ['vllm', 'serve', '/model', '--port', '80', '--served-model-name', 'llm', '--max-model-len', '4096', '--enforce-eager', '--trust-remote-code', '-tp', '1']  
+"""
+    
+    else:
+        raise ValueError(f"未知的 GPU_TYPE: {gpu_type}")
+    
+    
+    
+    
+
+# ══════════════════════════════════════════════════════════
+#  业务逻辑
+# ══════════════════════════════════════════════════════════
+# 账号"等待中/运行中"任务数已达上限时平台返回的业务错误信息（子串匹配）；
+# 命中这个的模型不算永久失败，会在额度腾出空位后自动重试，不会被记作 failed
+QUOTA_FULL_MSG = "当前等待中或运行中的异步模型验证任务数量已达上限"
+# 额度耗尽后，隔多久自动重试一次剩余（因额度问题未提交成功）的模型
+RETRY_INTERVAL_SECONDS = 30 * 60  # 30 分钟
+
+
+def _submit_task(token: str, gpu_type: str, model_id: str) -> Tuple[bool, str, str]:
+    headers = {
+        "Content-Type": "application/json",
+        "Authorization": f"Bearer {token}",
+    }
+    config_content = build_config_content(gpu_type, model_id)
+
     payload = {
         "contestApiToken": CONTEST_API_TOKEN,
         "contributors":    CONTRIBUTORS,
-        "gpuTypes":        [GPU_TYPE],
+        "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,
+            "gpuType":  gpu_type,
             "taskType": TASK_TYPE,
         }],
     }
+    print(f"[payload] gpu={gpu_type} model={model_id}", flush=True)
     try:
         resp = requests.post(
             BASE_URL + SUBMIT_ENDPOINT,
@@ -177,50 +376,88 @@ ref_config:
         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)
+            print(f"[worker] OK  {model_id} (GPU={gpu_type})  task_id={task_id}", flush=True)
-            return True, task_id
+            return True, task_id, ""
         else:
-            print(f"[worker] FAIL {model_id}: {result.get('message')}", flush=True)
+            message = result.get("message") or ""
-            return False, ""
+            print(f"[worker] FAIL {model_id} (GPU={gpu_type}): {message}", flush=True)
+            return False, "", message
     except Exception as e:
-        print(f"[worker] ERROR {model_id}: {e}", flush=True)
+        print(f"[worker] ERROR {model_id} (GPU={gpu_type}): {e}", flush=True)
-        return False, ""
+        return False, "", str(e)
 
 
 def _run_worker():
     _state["started_at"] = datetime.utcnow().isoformat()
     _state["phase"] = "submitting"
 
-    successful: List[Tuple[str, str]] = []
+    successful: List[Tuple[str, str, str]] = []
-    try:
+    token = AUTH_TOKEN
-        token = _login()
+    print("[worker] 使用预设 Token，跳过登录", flush=True)
-    except Exception:
-        traceback.print_exc()
-        _state["phase"] = "error"
-        return
 
-    for model_id in ALL_MODEL_IDS:
+    # 待提交队列：保持 GPU_JOBS 里原有的 (gpu_type, model_id) 顺序
-        if _shutdown.is_set():
+    pending: List[Tuple[str, str]] = [
-            break
+        (gpu_type, model_id)
-        ok, task_id = _submit_task(token, model_id)
+        for gpu_type, model_list in GPU_JOBS
-        if ok:
+        for model_id in model_list
-            _state["submitted"] += 1
+    ]
-            successful.append((task_id, model_id))
-        else:
-            _state["failed"] += 1
 
-    # 写入结果文件
+    round_num = 0
-    try:
+    while pending and not _shutdown.is_set():
-        with open("submitted_validation_tasks.txt", "w", encoding="utf-8") as f:
+        round_num += 1
-            for tid, mid in successful:
+        _state["round"] = round_num
-                f.write(f"{tid}\t{mid}\n")
+        _state["phase"] = "submitting"
-    except Exception:
+        _state["next_retry_at"] = None
-        pass
+        print(
+            f"\n{'='*60}\n🚀 第 {round_num} 轮，待提交 {len(pending)} 个模型\n{'='*60}",
+            flush=True,
+        )
+
+        quota_blocked: List[Tuple[str, str]] = []
+        for gpu_type, model_id in pending:
+            if _shutdown.is_set():
+                break
+            ok, task_id, message = _submit_task(token, gpu_type, model_id)
+            if ok:
+                _state["submitted"] += 1
+                _state["per_gpu"][gpu_type] += 1
+                successful.append((task_id, gpu_type, model_id))
+            elif QUOTA_FULL_MSG in message:
+                # 账号额度暂时满了，不算永久失败，留到下一轮重试
+                quota_blocked.append((gpu_type, model_id))
+            else:
+                # 非额度原因失败（如重复提交等），不再重试
+                _state["failed"] += 1
+
+        pending = quota_blocked
+        _state["quota_blocked_remaining"] = len(pending)
+
+        # 每轮结束都把已成功的结果落盘一次，避免中途重启丢失记录
+        try:
+            with open("submitted_validation_tasks.txt", "w", encoding="utf-8") as f:
+                for tid, gpu, mid in successful:
+                    f.write(f"{tid}\t{gpu}\t{mid}\n")
+        except Exception:
+            pass
+
+        if pending and not _shutdown.is_set():
+            next_retry = datetime.utcnow().timestamp() + RETRY_INTERVAL_SECONDS
+            _state["next_retry_at"] = datetime.utcfromtimestamp(next_retry).isoformat()
+            _state["phase"] = "waiting_retry"
+            print(
+                f"[worker] 第 {round_num} 轮结束：{len(pending)} 个模型因账号额度上限暂未提交，"
+                f"{RETRY_INTERVAL_SECONDS // 60} 分钟后自动重试（不部署新策略，本进程内循环）...",
+                flush=True,
+            )
+            _shutdown.wait(RETRY_INTERVAL_SECONDS)
 
     _state["finished_at"] = datetime.utcnow().isoformat()
     _state["phase"] = "done"
+    _state["quota_blocked_remaining"] = len(pending)
     print(
-        f"[worker] 完成  submitted={_state['submitted']}  failed={_state['failed']}",
+        f"[worker] 完成  submitted={_state['submitted']}  failed={_state['failed']}  "
+        f"total={_state['total']}  per_gpu={_state['per_gpu']}  "
+        f"仍因额度未提交(如遇shutdown中断)={len(pending)}",
         flush=True,
     )
     # 提交完成后继续保持进程存活，等待平台停止
@@ -253,4 +490,4 @@ def main():
 
 
 if __name__ == "__main__":
-    main()
+    main()

model_ids/hygon_k100-ai_2026-09-24.txt

@@ -0,0 +1,321 @@
+kaividlabs/qwen3-4b-awq
+glins7/cashgo-role_classification
+danielsaggau/scotus_f1
+arinze/address-match-abp-v5
+arinze/address-match-abp-v4
+danielsaggau/scotus_py
+yzhou286/mBio-finetuned-setfit-model
+Watwat100/140data
+Watwat100/256data
+Nhat1904/7_shot_STA_freezed_body_1e-5_batch_8
+Nhat1904/9_shot_STA_head_skhead_1epoch_16batch
+Nhat1904/10_shot_STA_head_skhead_1epoch
+Nhat1904/12_shot_STA_head_skhead
+yzhou286/mBio-setfit-model
+kanixwang/my-awesome-setfit-model
+futuredatascience/to-classifier-v1
+futuredatascience/from-classifier-v1
+kowshik/upsc-classification-model-v1
+lewispons/large-email-classifier
+IsaacRodgz/setfit-diff-head-stance-prediction-spanish-news-headlines
+gmsarti/setfit-ethos-multilabel-example
+YouLiXiya/tinyllava-v1.0-1.1b-hf
+javiervela/sentence-transformers_distiluse-base-multilingual-cased-v2_50-50_all-v2_sequence_oaei_final
+Vishwas/intent_classification
+airnicco8/xlm-roberta-en-it-de
+tubyneto/wandss-bert
+nategro/nps-mpnet
+nategro/nps-mpnet-lds
+mrm8488/setfit-mpnet-base-v2-finetuned-spam-detection
+kornwtp/ConGen-RoBERTa-base
+mencosk/Qwen2.5-Coder-1.5B-golang
+kornwtp/ConGen-TinyBERT-L6
+kornwtp/ConGen-BERT-Small
+kornwtp/ConGen-TinyBERT-L4
+kornwtp/ConGen-BERT-Mini
+kornwtp/ConGen-BERT-Tiny
+BlackKakapo/stsb-xlm-r-multilingual-ro
+TingChenChang/hpv-multi-qa-mpnet-zh
+Charul/my-dummy-model-1
+tubyneto/my_new_model
+TingChenChang/cMedQA2-multi-qa-mpnet-zh
+dvilasuero/my-test-setfit
+dvilasuero/setfit-mini-imdb
+mrm8488/setfit-mpnet-base-v2-finetuned-sentEval-CR
+bongsoo/moco-sentencedistilbertV2.1
+jamescalam/mpnet-nli-sts
+TingChenChang/hpvqa-lcqmc-ocnli-cnsd-multi-MiniLM-v2
+TingChenChang/lcqmc-ocnli-cnsd-multi-MiniLM-v2
+jamescalam/mpnet-snli-negatives
+teven/cross_all-mpnet-base-v2_finetuned_WebNLG2020_metric_average
+budecosystem/boomer-1b
+CShorten/CORD-19-Title-Abstracts-1-more-epoch
+teven/cross_all-mpnet-base-v2_finetuned_WebNLG2020_data_coverage
+valhalla/distilbart-mnli-12-9
+teven/bi_all-mpnet-base-v2_finetuned_WebNLG2020_data_coverage
+teven/cross_all-mpnet-base-v2_finetuned_WebNLG2020_correctness
+teven/bi_all-mpnet-base-v2_finetuned_WebNLG2020_correctness
+jamescalam/mpnet-snli
+jamescalam/mpnet-xnli
+teven/bi_all-mpnet-base-v2_finetuned_WebNLG2017
+jamescalam/mpnet-qa
+jamescalam/deberta-v3-base-qa
+firqaaa/indo-sentence-bert-base
+valhalla/distilbart-mnli-12-3
+rufimelo/Legal-BERTimbau-sts-base-ma-v2
+bongsoo/moco-sentencebertV2.0
+smartmind/roberta-ko-small-tsdae
+aiknowyou/all-mpnet-base-questions-clustering-en
+lewtun/dummy-setfit-model
+TingChenChang/qqp-nli-training-paraphrase-multilingual-MiniLM-L12-v2
+bongsoo/moco-sentencedistilbertV2.0
+edumunozsala/bertin-sts-cc-news-es
+edumunozsala/distilroberta-sentence-transformer-test
+mchochlov/codebert-base-cd-ft
+smartmind/ko-sbert-augSTS-maxlength512
+lmxhappy/yule_bagua_bert
+spacemanidol/esci-all-distilbert-base-uncased-5e-5
+intfloat/simlm-msmarco-reranker
+AI-Growth-Lab/Pap2PatentSBERTa
+Kyleiwaniec/COS_TAPT_n_RoBERTa_STS
+spacemanidol/esci-es-mpnet-crossencoder
+spacemanidol/esci-jp-mpnet-crossencoder
+spacemanidol/esci-mpnet-crossencoder
+osanseviero/distilroberta-base-sentence-transformer
+embedding-data/distilroberta-base-sentence-transformer
+embedding-data/deberta-sentence-transformer
+ivan-savchuk/msmarco-distilbert-dot-v5-tuned-full-v1
+sdadas/st-polish-paraphrase-from-mpnet
+sdadas/st-polish-paraphrase-from-distilroberta
+Daveee/gpl_colbert
+sorayutmild/simcse-model-wangchanberta-finetuned-sanook-news
+CaoHaiNam/vietnamese-address-embedding
+aiknowyou/aiky-sentence-bertino
+TimKond/S-BioLinkBert-MedQuAD
+NimaBoscarino/STPushToHub-test
+NimaBoscarino/albert-nima
+alfaneo/bertimbaulaw-base-portuguese-sts
+alfaneo/jurisbert-base-portuguese-sts
+alfaneo/bertimbau-base-portuguese-sts
+alfaneo/bert-base-multilingual-sts
+WalidLak/Testmodel
+shafin/distilbert-similarity-b32-3
+raphaelsty/semanlink_all_mpnet_base_v2
+guidecare/all-mpnet-base-v2-feature-extraction
+income/bpr-gpl-climate-fever-base-msmarco-distilbert-tas-b
+income/bpr-gpl-dbpedia-entity-base-msmarco-distilbert-tas-b
+income/bpr-gpl-hotpotqa-base-msmarco-distilbert-tas-b
+income/bpr-gpl-nfcorpus-base-msmarco-distilbert-tas-b
+income/bpr-gpl-scifact-base-msmarco-distilbert-tas-b
+income/bpr-gpl-trec-covid-base-msmarco-distilbert-tas-b
+income/bpr-gpl-trec-news-base-msmarco-distilbert-tas-b
+income/bpr-gpl-webis-touche2020-base-msmarco-distilbert-tas-b
+ITESM/sentece-embeddings-BETO
+espejelomar/sentece-embeddings-BETO
+xverse/XVERSE-13B-Chat
+ceggian/sbert_pt_reddit_mnr_128
+ceggian/sbert_pt_reddit_mnr_256
+ceggian/sbert_pt_reddit_softmax_512
+orenpereg/paraphrase-mpnet-base-v2_sst2_64samps
+ceggian/sbert_pt_reddit_mnr_512
+orenpereg/paraphrase-mpnet-base-v2_sst2_4samps
+GPL/bioasq-msmarco-distilbert-gpl
+GPL/bioasq-tsdae-msmarco-distilbert-gpl
+GPL/scidocs-tsdae-msmarco-distilbert-gpl
+GPL/quora-tsdae-msmarco-distilbert-gpl
+GPL/nfcorpus-tsdae-msmarco-distilbert-gpl
+GPL/dbpedia-entity-tsdae-msmarco-distilbert-gpl
+GPL/hotpotqa-msmarco-distilbert-gpl
+GPL/quora-distilbert-tas-b-gpl-self_miner
+GPL/hotpotqa-distilbert-tas-b-gpl-self_miner
+kevinpro/MetaMathOctopus-MAPO-DPO-13B
+laion/exp-syh-r2egym-swesmith-mixed_glm_4_7_traces_jupiter_cleaned
+ceggian/sbert_standard_reddit_mnr
+snunlp/KR-SBERT-V40K-klueNLI-augSTS
+deepset/all-mpnet-base-v2-table
+GPL/trec-news-tsdae-msmarco-distilbert-gpl
+ml6team/cross-encoder-mmarco-german-distilbert-base
+GPL/fever-tsdae-msmarco-distilbert-gpl
+GPL/nfcorpus-msmarco-distilbert-gpl
+GPL/dbpedia-entity-msmarco-distilbert-gpl
+efederici/sentence-BERTino
+efederici/sentence-bert-base
+mrp/SimCSE-model-WangchanBERTa-V2
+GPL/scifact-distilbert-tas-b-gpl-self_miner
+sentence-transformers/stsb-bert-large
+sentence-transformers/stsb-bert-base
+sentence-transformers/sentence-t5-large
+sentence-transformers/quora-distilbert-multilingual
+GPL/climate-fever-tsdae-msmarco-distilbert-gpl
+GPL/arguana-tsdae-msmarco-distilbert-gpl
+GPL/trec-covid-msmarco-distilbert-gpl
+GPL/scidocs-msmarco-distilbert-gpl
+GPL/webis-touche2020-msmarco-distilbert-gpl
+GPL/trec-news-msmarco-distilbert-gpl
+GPL/signal1m-msmarco-distilbert-gpl
+GPL/quora-msmarco-distilbert-gpl
+GPL/nq-msmarco-distilbert-gpl
+GPL/climate-fever-msmarco-distilbert-gpl
+GPL/newsqa-msmarco-distilbert-gpl
+ddobokki/unsup-simcse-klue-roberta-small
+GPL/arguana-msmarco-distilbert-gpl
+sentence-transformers/use-cmlm-multilingual
+jegormeister/robbert-v2-dutch-base-mqa-finetuned
+meedan/paraphrase-filipino-mpnet-base-v2
+bespin-global/klue-sroberta-base-continue-learning-by-mnr
+DMetaSoul/sbert-chinese-qmc-domain-v1-distill
+somosnlp-hackathon-2022/paraphrase-spanish-distilroberta
+somosnlp-hackathon-2022/bertin-roberta-base-finetuning-esnli
+DMetaSoul/sbert-chinese-qmc-finance-v1
+sentence-transformers/xlm-r-base-en-ko-nli-ststb
+sentence-transformers/nli-distilbert-base-max-pooling
+NastasiaM/mbert-loraxs-qa-vanilla
+sentence-transformers/multi-qa-mpnet-base-dot-v1
+sentence-transformers/msmarco-distilbert-base-v4
+morethankk/ThermalGuard-v1_4
+DMetaSoul/sbert-chinese-dtm-domain-v1
+DMetaSoul/sbert-chinese-qmc-domain-v1
+DMetaSoul/sbert-chinese-general-v1
+moshew/paraphrase-mpnet-base-v2_SetFit_sst2
+mariolux/sherpa-onnx-whisper-tiny
+mariolux/sherpa-onnx-whisper-small
+mariolux/sherpa-onnx-telespeech-ctc-zh-2024-06-04
+whaleloops/phrase-bert
+mariolux/sherpa-onnx-fire-red-asr-large-zh_en-2025-02-16
+lzkhhh/ITDR-Qwen2.5-7B
+longvideotool/LongVT-SFT
+valurank/paraphrase-mpnet-base-v2-offensive
+usc-isi/sbert-roberta-large-anli-mnli-snli
+uer/sbert-base-chinese-nli
+symanto/sn-xlm-roberta-base-snli-mnli-anli-xnli
+symanto/sn-mpnet-base-snli-mnli
+lm2445/TABPO_llama3.1_8B_3epoch
+sentence-transformers/xlm-r-large-en-ko-nli-ststb
+sentence-transformers/xlm-r-distilroberta-base-paraphrase-v1
+sentence-transformers/xlm-r-bert-base-nli-stsb-mean-tokens
+sentence-transformers/xlm-r-bert-base-nli-mean-tokens
+sentence-transformers/xlm-r-100langs-bert-base-nli-stsb-mean-tokens
+sentence-transformers/xlm-r-100langs-bert-base-nli-mean-tokens
+sentence-transformers/stsb-xlm-r-multilingual
+sentence-transformers/stsb-roberta-large
+sentence-transformers/stsb-mpnet-base-v2
+sentence-transformers/sentence-t5-xxl
+sentence-transformers/sentence-t5-base
+sentence-transformers/roberta-large-nli-mean-tokens
+sentence-transformers/paraphrase-mpnet-base-v2
+sentence-transformers/paraphrase-TinyBERT-L6-v2
+sentence-transformers/paraphrase-MiniLM-L6-v2
+sentence-transformers/paraphrase-MiniLM-L3-v2
+sentence-transformers/paraphrase-MiniLM-L12-v2
+sentence-transformers/nq-distilbert-base-v1
+sentence-transformers/nli-roberta-large
+sentence-transformers/nli-roberta-base
+sentence-transformers/nli-roberta-base-v2
+sentence-transformers/nli-mpnet-base-v2
+sentence-transformers/nli-distilroberta-base-v2
+iic/speech_conformer_asr_nat-zh-cn-16k-aishell1-vocab4234-pytorch
+iic/speech_UniASR_asr_2pass-zh-cn-16k-common-vocab8358-tensorflow1-online
+iic/speech_UniASR_asr_2pass-ru-16k-common-vocab1664-tensorflow1-online
+iic/speech_UniASR_asr_2pass-id-16k-common-vocab1067-tensorflow1-online
+iic/speech_UniASR_asr_2pass-he-16k-common-vocab1085-pytorch
+iic/speech_UniASR_asr_2pass-en-16k-common-vocab1080-tensorflow1-online
+iic/speech_UniASR_asr_2pass-cantonese-CHS-16k-common-vocab1468-tensorflow1-online
+sentence-transformers/nli-distilbert-base
+sentence-transformers/nli-bert-large
+sentence-transformers/multi-qa-mpnet-base-cos-v1
+sentence-transformers/gtr-t5-xxl
+sentence-transformers/gtr-t5-xl
+dengcunqin/speech_seaco_paraformer_large_asr_nat-zh-cantonese-en-16k-common-vocab11666-pytorch
+sentence-transformers/nli-bert-large-cls-pooling
+Mozilla/llava-v1.5-7b-llamafile
+sentence-transformers/nli-bert-base-cls-pooling
+sentence-transformers/multi-qa-distilbert-dot-v1
+sentence-transformers/multi-qa-distilbert-cos-v1
+sentence-transformers/multi-qa-MiniLM-L6-dot-v1
+sentence-transformers/multi-qa-MiniLM-L6-cos-v1
+sentence-transformers/msmarco-distilbert-dot-v5
+sentence-transformers/msmarco-distilbert-cos-v5
+sentence-transformers/msmarco-distilbert-base-v2
+sentence-transformers/msmarco-distilbert-base-dot-prod-v3
+sentence-transformers/msmarco-bert-co-condensor
+sentence-transformers/msmarco-bert-base-dot-v5
+sentence-transformers/msmarco-MiniLM-L12-cos-v5
+youngfficy/feifei-qwen2.5-1.5b-catgirl
+sentence-transformers/msmarco-MiniLM-L6-v3
+q2792046875/internVL1B
+muse/openai-clip-vit-large-patch14
+sentence-transformers/facebook-dpr-question_encoder-single-nq-base
+sentence-transformers/facebook-dpr-question_encoder-multiset-base
+laion/Qwen3-8B_exp_tas_top_k_32_traces_save-strategy_steps
+sentence-transformers/facebook-dpr-ctx_encoder-multiset-base
+sentence-transformers/distiluse-base-multilingual-cased-v1
+sentence-transformers/distilroberta-base-msmarco-v1
+sentence-transformers/distilroberta-base-msmarco-v2
+sentence-transformers/distilbert-base-nli-mean-tokens
+laion/GLM-4_7-stackexchange-tezos-sandboxes-maxeps-131k
+Vchitect/ShotVL-3B
+OpenGVLab/VideoChat-R1_7B
+sentence-transformers/all-mpnet-base-v1
+sentence-transformers/all-MiniLM-L6-v1
+tsss1/deepsek-qwen1.5-vpn
+reedmayhew/gemma3-12B-claude-3.7-sonnet-reasoning-distilled
+starVLA/Qwen3-VL-4B-Instruct-Action
+mistralai/Pixtral-12B-2409
+mlfoundations-cua-dev/qwen2_5vl_7b_easyr1_10k_hard_qwen7b_easy_gta17b_or_segui3b-4MP
+mlfoundations-cua-dev/qwen2_5vl_7b_easyr1_10k_hard_segui3b_easy_gta1-4MP
+mlfoundations-cua-dev/qwen2_5vl_3b_sft_idm_how_to_onannel_agent_sft_data_local_bs_4_epochs_3
+ibm-granite/granite-4.1-30b
+osanseviero/clip-st
+new5558/simcse-model-wangchanberta-base-att-spm-uncased
+navteca/multi-qa-mpnet-base-cos-v1
+navteca/all-mpnet-base-v2
+nanopass/test-model-fe
+mrp/simcse-model-m-bert-thai-cased
+mrm8488/roberta-base-bne-finetuned-sqac-retriever
+ncls-p/Qwen2.5-7B-blog-key-points
+laion/openthoughts-4-code-qwen3-32b-annotated-32k_qwen3-1.7B_32k
+aab20abdullah/qwen_OSINT
+OpenGVLab/InternVL3-1B-Instruct
+sakares/wav2vec2-large-xlsr-thai-demo
+CuongLD/wav2vec2-large-xlsr-vietnamese
+cahya/wav2vec2-large-xlsr-indonesian
+indonesian-nlp/wav2vec2-large-xlsr-indonesian-baseline
+indonesian-nlp/wav2vec2-large-xlsr-indonesian
+m3hrdadfi/wav2vec2-large-xlsr-persian-v3
+nguyenvulebinh/wav2vec2-base-vietnamese-250h
+airesearch/wav2vec2-large-xlsr-53-th
+indonesian-nlp/wav2vec2-indonesian-javanese-sundanese
+ctl/wav2vec2-large-xlsr-cantonese
+jonatasgrosman/wav2vec2-large-xlsr-53-persian
+jonatasgrosman/wav2vec2-large-xlsr-53-arabic
+muzamil47/wav2vec2-large-xlsr-53-arabic-demo
+waltonfuture/qwen2.5vl-3b-sampled_5000_qwen2.5vl32b
+waltonfuture/qwen2.5vl-3b-32b-longest-5153
+RedHatAI/Qwen2.5-VL-3B-Instruct-quantized.w8a8
+kresnik/wav2vec2-large-xlsr-korean
+mlfoundations-cua-dev/qwen2_5vl_3b_sft_unified_idm_data_with_new_idm_data_2_frames_local_bs_1
+OpenMed/Qwen2.5-3B-MedVL
+maxidl/wav2vec2-large-xlsr-german
+imvladikon/wav2vec2-xls-r-300m-hebrew
+dbdmg/wav2vec2-xls-r-300m-italian-robust
+mikr/whisper-large-v3-czech-cv13
+ocordeiro/w2v-bert-2.0-portuguese-colab-CV16.0
+whitefox123/w2v-bert-2.0-arabic-4
+01ai/Yi-VL-6B
+jerchenxin/qwen2.5-Math-1.5B-step-720
+jerchenxin/qwen2.5-Math-1.5B-step-320
+KandirResearch/DarijaTTS-v0.1-500M
+allura-org/remnant-qwen3-8b
+bimabk/test_19fccc14-8df6-4085-86ee-ce740ccdff30_JackFram-llama-160m
+bimabk/test_19fccc14-8df6-4085-86ee-ce740ccdff30_Qwen-Qwen1-5-0-5B-Chat
+bimabk/test_19fccc14-8df6-4085-86ee-ce740ccdff30_unsloth-Qwen2-0-5B
+bimabk/test_19fccc14-8df6-4085-86ee-ce740ccdff30_Qwen-Qwen2-5-0-5B
+bimabk/test_19fccc14-8df6-4085-86ee-ce740ccdff30_TinyLlama-TinyLlama-1-1B-Chat-v0-6
+gaoqie/Qwen2VL-2B-Instruct-fire
+bimabk/test_ac92fa52-28b8-479a-b5d5-a678407b5011_unsloth-Qwen2-5-3B
+bimabk/test_ac92fa52-28b8-479a-b5d5-a678407b5011_Qwen-Qwen2-5-3B-Instruct
+qingy2024/Benchmaxx-Llama-3.2-1B-Instruct
+unsloth/orpheus-3b-0.1-ft
+diabolic6045/Sanskrit-qwen-7B-Translate-v2
+ayoubkirouane/whisper-small-ar
+SEGAgentRL/LLDS-A-GSPO-Qwen2.5-3B-Ins

model_ids/ppu_zw_810e_2026-09-24.txt

@@ -0,0 +1,17 @@
+Xlnk/LFM2-2.6B-Exp-GGuf
+internlm/internlm2-7b-reward
+jbuaba/iolai-2026-qwen25-14b
+KBlueLeaf/TIPOv2-1B-A200M
+glins7/cashgo-role_classification
+danielsaggau/scotus_f1
+arinze/address-match-abp-v5
+arinze/address-match-abp-v4
+Nhat1904/10_shot_STA_head_skhead_1epoch
+Nhat1904/12_shot_STA_head_skhead
+Nhat1904/4_shot_STA
+shrinivasbjoshi/setfit-mbti-multiclass-w266_Nov29
+Etelis/rtm_fewshot
+TheDrummer/Snowpiercer-15B-v4
+gaunernst/gemma-3-27b-it-qat-autoawq
+darkps/darkit-v1.5
+EleutherAI/pythia-70m