submit 24 models on zhoushasha: ppu_zw_810e(18)/hygon_k100-ai(4)/MetaX_c-500(1)/Biren_166m(1)

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>

main.py

@@ -46,257 +46,61 @@ HTTP_PORT = 8080
 #  各 GPU 的模型列表（来自 filter_verified_models 脚本的筛选结果）
 # ══════════════════════════════════════════════════════════
 BIREN_MODELS = [
-    "ApolloRaines/Phi-4-mini-Instruct-Desyced",
-    "allenai/OLMo-2-0425-1B",
-    "Free2035/4QDR_4B_AD_Thinker_V1",
-    "aifoundry-org/OLMo-7B-0424-hf-Quantized",
-    "lugman-madhiai/Qwen3-4B-MHS-1.1",
-    "zypchn/BehChat-SFT-v4",
-    "NoesisLab/Kai-30B-Instruct",
-    "barandinho/Qwen3-30B-A3B-FIRST-STAGE-SFT",
-    "m-a-p/OpenLLaMA-Reproduce-218.1B",
-    "xiaolesu/Qwen3-8B-Herald-SFT",
-    "WhiteRabbitNeo/WhiteRabbitNeo-33B-v1.5",
-    "saleh1312/orph_3.07225",
-    "vanta-research/atom-olmo3-7b",
-    "ZhipuAI/LongCite-glm4-9b",
-    "Xlnk/LFM2-2.6B-Exp-GGuf",
-    "Tesslate/UIGEN-T1.1-Qwen-14B",
-    "jondurbin/bagel-dpo-34b-v0.2",
-    "zhengr/MixTAO-7Bx2-MoE-Instruct-v5.0",
-    "allenai/Olmo-3.1-32B-Instruct",
-    "bjaidi/Phi-3-medium-128k-instruct-awq",
-    "xing720310/qwen3-14b",
-    "IntelLabs/sqft-mistral-7b-v0.3-50-base-gptq",
-    "hariharanv04/qwen2.5-coder-14b-metadata-merged",
-    "junfengzhou/qwen3-14b-rl",
-    "ronnywebdevs1/Affine-P011-5CkU7wLMWXPs6TdSsMf8eEYCVAbPLyNmYg9PPx1Uds8toKra",
-    "kennedyantonio0301/Affine-Tensor-h3-5EkdoaCmEpFffUjDpLhDMzEDR4kptaEzpTPYCP1uL2sbct8C",
-    "julep-ai/dolphin-2.9-llama3-70b-awq",
-    "cortexso/gemma3",
-    "jacob-ml/jacob-24b",
-    "lyraaaa/neuralese-sft-pretrain-v2",
-    "ai-sage/GigaChat-20B-A3B-instruct-bf16",
-    "PJMixers-Dev/gemma-3-1b-it-fixed",
-    "commotion/svara_finetune_v1",
-    "geoffmunn/Qwen3-14B-f16",
-    "geoffmunn/Qwen3-32B-f16",
-    "TeichAI/Nemotron-Cascade-14B-Thinking-Claude-4.5-Opus-High-Reasoning-Distill",
-    "prithivMLmods-llamafile/SmolLM2-1.7B-Instruct-llamafile",
-    "prithivMLmods-llamafile/Llama-3.2-8B-llamafile-200K",
-    "llamafile-club/SmolLM-135M-Instruct-Llamafile",
-    "prithivMLmods/Sombrero-QwQ-32B-Elite9",
-    "prithivMLmods-llamafile/Aya-Expanse-8B-llamafile",
-    "llamafile-club/SmolLM-135M-Llamafile",
-    "prithivMLmods/Sombrero-QwQ-32B-Elite10-Fixed",
-    "prithivMLmods-llamafile/Qwen2.5-Coder-1.5B-llamafile",
-    "TeichAI/Qwen3-14B-Polaris-Alpha-Distill",
-    "okwinds/MiroThinker-14B-DPO-v0.1",
-    "sanbuphy/tianji-wish2-14b",
-    "codefuse-ai/CodeFuse-StarCoder2-15B",
-    "AI-ModelScope/txgemma-27b-chat",
-    "Shanghai_AI_Laboratory/internlm3-8b-instruct-awq",
-    "XGenerationLab/XiYanSQL-QwenCoder-32B-2412",
-    "vllm-ascend/gemma-1.1-2b-it",
-    "OpenBuddy/openbuddy-qwen1.5-32b-v21.2-32k",
-    "OpenBuddy/openbuddy-thinker-32b-v26-preview",
-    "OpenBuddy/openbuddy-qwen1.5-32b-v21.1-32k",
-    "TechxGenus-MS/CodeGemma-7b",
-    "OpenBuddy/openbuddy-qwq-32b-v25.2q-200k",
-    "unsloth/Qwen3-30B-A3B",
-    "OpenBuddy/openbuddy-qwq-32b-v25.1-200k",
-    "OpenBuddy/openbuddy-r1-32b-v24.1-200k",
-    "iic/ERank-14B",
-    "OpenBuddy/openbuddy-yi1.5-34b-v21.2-32k",
-    "LGAI-EXAONE/EXAONE-Deep-32B",
-    "OpenBuddy/openbuddy-qwq-32b-v24.2-200k",
-    "unsloth/Phi-3-mini-4k-instruct-v0",
-    "argilla/notux-8x7b-v1",
-    "voidful/qd-phi-1_5",
-    "Shanghai_AI_Laboratory/internlm2-math-base-20b",
-    "Shanghai_AI_Laboratory/internlm2-math-plus-20b",
-    "TechxGenus-MS/starcoder2-15b-instruct",
-    "Shanghai_AI_Laboratory/internlm2-base-20b",
-    "m-a-p/OpenLLaMA-Reproduce-872.42B",
-    "m-a-p/OpenLLaMA-Reproduce-973.08B",
-    "Shanghai_AI_Laboratory/OREAL-32B",
-    "YOYO-AI/Qwen3-30B-A3B-CoderThinking-YOYO-linear",
-    "ticoAg/Qwen-1_8B-Chat-Int4-awq",
-    "smirki/UIGEN-T1.1-Qwen-14B",
-    "prithivMLmods/Qwen2.5-32B-DeepSeek-R1-Instruct",
-    "sail/Sailor2-20B-128K",
-    "xverse/XVERSE-65B",
-    "Shanghai_AI_Laboratory/internlm2_5-20b-chat",
-    "TeleAI/TeleChat-52B",
-    "modelscope/Llama-2-70b-ms",
-    "Shanghai_AI_Laboratory/internlm2-20b",
-    "ai-modelscope/Llama-3_1-Nemotron-51B-Instruct",
-    "zhuangxialie/Phi-3-Chinese-ORPO",
-    "openai-mirror/gpt-oss-safeguard-20b",
-    "ByteDance-Seed/Seed-OSS-36B-Instruct",
-    "Shanghai_AI_Laboratory/internlm-chat-20b",
-    "TurkuNLP/bloom-finnish-176b",
-    "openai-mirror/gpt-oss-120b",
-    "vllm-ascend/QwQ-32B-W8A8",
-    "mistralai/Mistral-Small-24B-Instruct-2501",
-    "Shanghai_AI_Laboratory/internlm-20b",
-    "ZhipuAI/GLM-4-32B-0414",
+    "zipaltrivedi/dotnet-coder-14b",
 ]
 
 CAMBRICON_MODELS = [
-    "Xlnk/LFM2-2.6B-Exp-GGuf",
 ]
 
 METAX_MODELS = [
-    "ApolloRaines/Phi-4-mini-Instruct-Desyced",
-    "robertspumiaca1975/Qwen2.5-Coder-14B-n8n-Workflow-Generator",
-    "barandinho/Qwen3-30B-A3B-FIRST-STAGE-SFT-V2",
-    "hotmailuser/QwenSlerp2-14B",
-    "WhiteRabbitNeo/WhiteRabbitNeo-33B-v1.5",
-    "madox81/SmolLM2-135M-cybersecurity-lora-merged",
-    "Xlnk/LFM2-2.6B-Exp-GGuf",
-    "Tesslate/UIGEN-T1.1-Qwen-14B",
-    "jondurbin/bagel-dpo-34b-v0.2",
-    "zhengr/MixTAO-7Bx2-MoE-Instruct-v5.0",
-    "bjaidi/Phi-3-medium-128k-instruct-awq",
-    "jacob-ml/jacob-24b",
-    "geoffmunn/Qwen3-32B-f16",
-    "TorpedoSoftware/Luau-Devstral-24B-Instruct-v0.2",
-    "TeichAI/Nemotron-Cascade-14B-Thinking-Claude-4.5-Opus-High-Reasoning-Distill",
-    "tongzang/Qwen2.5-7b-lora-law",
-    "prithivMLmods/Sombrero-QwQ-32B-Elite9",
-    "prithivMLmods/Sombrero-QwQ-32B-Elite10-Fixed",
-    "TeichAI/Qwen3-14B-Polaris-Alpha-Distill",
-    "okwinds/MiroThinker-14B-DPO-v0.1",
-    "sanbuphy/tianji-wish2-14b",
-    "YOYO-AI/YOYO-O1-14B",
-    "LLM-Research/Meta-Llama-3.1-405B",
-    "LLM-Research/Meta-Llama-3-70B",
-    "LLM-Research/Meta-Llama-3.1-70B",
-    "Qwen/Qwen-72B-Chat",
-    "Qwen/Qwen3-Coder-480B-A35B-Instruct",
-    "deepseek-ai/DeepSeek-R1-Distill-Llama-70B",
-    "codefuse-ai/CodeFuse-StarCoder2-15B",
-    "AI-ModelScope/txgemma-27b-chat",
-    "XGenerationLab/XiYanSQL-QwenCoder-32B-2412",
-    "OpenBuddy/openbuddy-qwen1.5-32b-v21.2-32k",
-    "OpenBuddy/openbuddy-thinker-32b-v26-preview",
-    "OpenBuddy/openbuddy-qwen1.5-32b-v21.1-32k",
-    "OpenBuddy/openbuddy-qwq-32b-v25.2q-200k",
-    "unsloth/Qwen3-30B-A3B",
-    "OpenBuddy/openbuddy-qwq-32b-v25.1-200k",
-    "OpenBuddy/openbuddy-r1-32b-v24.1-200k",
-    "iic/ERank-14B",
-    "OpenBuddy/openbuddy-yi1.5-34b-v21.2-32k",
-    "LGAI-EXAONE/EXAONE-Deep-32B",
-    "OpenBuddy/openbuddy-qwq-32b-v24.2-200k",
-    "unsloth/Phi-3-mini-4k-instruct-v0",
-    "argilla/notux-8x7b-v1",
-    "voidful/qd-phi-1_5",
-    "TechxGenus-MS/starcoder2-15b-instruct",
-    "m-a-p/OpenLLaMA-Reproduce-872.42B",
-    "m-a-p/OpenLLaMA-Reproduce-973.08B",
-    "Shanghai_AI_Laboratory/OREAL-32B",
-    "YOYO-AI/Qwen3-30B-A3B-CoderThinking-YOYO-linear",
-    "smirki/UIGEN-T1.1-Qwen-14B",
-    "sthenno-com/miscii-14b-0130",
-    "prithivMLmods/Qwen2.5-32B-DeepSeek-R1-Instruct",
-    "sail/Sailor2-20B-128K",
-    "xverse/XVERSE-65B",
-    "TeleAI/TeleChat-52B",
-    "modelscope/Llama-2-70b-ms",
-    "Shanghai_AI_Laboratory/internlm2-20b",
-    "ai-modelscope/Llama-3_1-Nemotron-51B-Instruct",
-    "aJupyter/EmoLLM_Qwen2-7B-Instruct_lora",
-    "zhuangxialie/Phi-3-Chinese-ORPO",
-    "vllm-ascend/QwQ-32B-W8A8",
-    "ZhipuAI/GLM-4-32B-0414",
+    "zipaltrivedi/dotnet-coder-14b",
 ]
 
 HYGON_MODELS = [
-    "ApolloRaines/Phi-4-mini-Instruct-Desyced",
+    "zipaltrivedi/dotnet-coder-14b",
+    "cds-jb/qwen3-14b-butterfly-subliminal-fullft",
+    "lllqaq/Qwen2.5-Coder-14B-Instruct-num11-v1-v2-v3-pairs-v3-triples-post-r2egym",
+    "deepmako/Mako-32B-Conductor",
 ]
 
 KUNLUNXIN_MODELS = [
-    "Xlnk/LFM2-2.6B-Exp-GGuf",
 ]
 
 PPU_MODELS = [
-    
-    "OuteAI/Lite-Mistral-150M-v2-Instruct",
-    "eric0009/yi-ko-6b-text2sql",
-    "ai-forever/mGPT-1.3B-bashkir",
-    "ApolloRaines/Phi-4-mini-Instruct-Desyced",
-    "eekay/Llama-3.1-8B-Instruct-noised-np0.15-emb-s48",
-    "eekay/Llama-3.1-8B-Instruct-noised-np0.15-emb-s49",
-    "cortexso/simplescaling-s1",
-    "BrainDelay/Siren",
-    "sbintuitions/sarashina2.2-3b-instruct-v0.1",
-    "robertspumiaca1975/Qwen2.5-Coder-14B-n8n-Workflow-Generator",
-    "AtAndDev/ShortKing-3b-v0.2",
-    "athirdpath/Iambe-RP-cDPO-20b",
-    "belweave/kai-2",
-    "unsloth/Qwen2.5-Coder-14B-Instruct",
-    "dphn/dolphin-2.7-mixtral-8x7b",
-    "adeljebali/llama3.1-gec-strict",
-    "xxrickyxx/Ailo152m-events-en",
-    "RedHatAI/starcoder2-7b-quantized.w8a8",
-    "RedHatAI/granite-3.1-2b-instruct-quantized.w4a16",
-    "julep-ai/dolphin-2.9.1-llama-3-70b-awq",
-    "OpenBuddy/openbuddy-deepseek-67b-v18.1-4k-gptq",
-    "dessertlab/offensive-powershell-CodeGPT-small",
-    "misterJB/atlas-field-528hz",
-    "tiiuae/Falcon3-10B-Base",
-    "Jackrong/gpt-oss-120b-Distill-Llama3.1-8B-v3",
-    "TheBloke/guanaco-65B-HF",
-    "jondurbin/airoboros-33b-gpt4-1.3",
-    "h2oai/h2ogpt-4096-llama2-70b",
-    "jondurbin/airoboros-65b-gpt4-1.3",
-    "jondurbin/airoboros-l2-70b-gpt4-2.0",
-    "ICBU-NPU/FashionGPT-70B-V1.2",
-    "jukofyork/Dark-Miqu-70B",
-    "alnrg2arg/blockchainlabs_joe_bez_seminar",
-    "facebook/opt-66b",
-    "abchbx/qwen_1.8B_Muice-Dataset_FULL",
-    "LumiOpen/Viking-33B",
-    "adamo1139/Yi-34B-200K-AEZAKMI-RAW-1701",
-    "mesolitica/Malaysian-TTS-4B-v0.1",
-    "TomGrc/FusionNet_passthrough",
-    "YOYO-AI/Qwen3-30B-A3B-YOYO-V5",
-    "m-a-p/OpenLLaMA-Reproduce-536.87B",
-    "m-a-p/OpenLLaMA-Reproduce-1291.85B",
-    "KnutJaegersberg/Deacon-34B",
-    "SenseLLM/ReflectionCoder-DS-33B",
-    "KOREAson/KO-REAson-AX3_1-35B-1009",
-    "dphn/dolphin-2.9.1-mixtral-1x22b",
-    "jondurbin/airoboros-33b-gpt4-1.4",
-    "TomGrc/FusionNet_passthrough_v0.1",
-    "Mozilla/Mistral-7B-Instruct-v0.2-llamafile",
-    "suayptalha/Luminis-phi-4",
-    "casperhansen/llama-3.3-70b-instruct-awq",
-    "HIT-SCIR/Chinese-Mixtral-8x7B",
-    "Shanghai_AI_Laboratory/internlm2-wqx-20b",
-    "unsloth/Qwen2.5-Coder-32B-Instruct",
-    "BSC-LT/ALIA-40b",
-    "Shanghai_AI_Laboratory/internlm2-7b",
-    "Shanghai_AI_Laboratory/internlm2-chat-7b",
-    
-    
-    
-    
+    "fpadovani/eng-latn-100mb-after-ppt-Dp-10mb-ckpt500_seed10",
+    "fpadovani/ita-latn-10mb-after-ppt-Dp-100mb-ckpt500_seed3407",
+    "fpadovani/ita-latn-10mb-after-ppt-Dp-10mb-ckpt500_seed3407",
+    "richardr1126/spider-skeleton-wizard-coder-merged",
+    "fpadovani/dan-latn-10mb-after-ppt-shuff-dyck-10mb-ckpt500_seed3407",
+    "fpadovani/ita-latn-10mb-after-ppt-shuff-dyck-100mb-ckpt500_seed3407",
+    "omerkaragulmez/XbyK-0.1",
+    "fpadovani/eng-latn-10mb-after-ppt-Dp-10mb-ckpt500_seed10",
+    "fpadovani/eng-latn-10mb-100mb_seed10",
+    "flax-community/gpt2-medium-indonesian",
+    "RedHatAI/QwQ-32B-Preview-quantized.w8a8",
+    "fractalego/fact-checking",
+    "KoboldAI/GPT-J-6B-Adventure",
+    "EasierAI/Falcon-3-1B",
+    "theprint/mistral-7b-cthulhu",
+    "iwalton3/phoenix",
+    "renzhenzhen/internLM2-for-triples",
+    "u2mithrandir/epsi_tmall",
 ]
 
-# 本次提交第二十四轮过滤结果 MetaX_c-500(63) / Kunlunxin_p-800(1) /
-# Cambricon_mlu-370-x8(1) / Biren_166m(95)，加上已更新的 ppu_zw_810e(57)，共217个。
-# hygon_k100-ai 本轮筛出1个，config 分支与 HYGON_MODELS 列表已备好但按要求不提交；
-# Iluvatar_bi-150 本轮筛出36个，本仓库 framework=vllm 而现有 iluvatar 镜像均为
-# llamacpp/GGUF，缺 vllm 版镜像地址，无 config 分支，同样不提交。
+# 本轮提交：ppu_zw_810e(18) / hygon_k100-ai(4) / MetaX_c-500(1) / Biren_166m(1)，共24个。
+#
+# 候选池是 api_verify_model_download_status_a.txt 里那 24711 个【已下载】的模型，
+# 正好满足机制A「模型必须已下载到平台存储」的前提（v1.0.37 那次157个失败就是因为没下载）。
+#
+# 本轮严格口径与放宽口径（别的卡「已验证」vs「已验证或验证中」）结果完全相同：
+# 这批已下载模型里 24007 个有验证记录的，24006 个都已至少一张卡「已验证」，
+# 放宽只多捞出 1 个；真正的瓶颈是各卡「无记录」的模型太少——
+# Kunlunxin_p-800 / Cambricon_mlu-370-x8 / Iluvatar_bi-150 均为 0，故不列入 GPU_JOBS。
 GPU_JOBS: List[Tuple[str, List[str]]] = [
-    ("MetaX_c-500",          METAX_MODELS),
-    ("Kunlunxin_p-800",      KUNLUNXIN_MODELS),
-    ("Cambricon_mlu-370-x8", CAMBRICON_MODELS),
-    ("Biren_166m",           BIREN_MODELS),
     ("ppu_zw_810e",          PPU_MODELS),
+    ("hygon_k100-ai",        HYGON_MODELS),
+    ("MetaX_c-500",          METAX_MODELS),
+    ("Biren_166m",           BIREN_MODELS),
 ]
 TOTAL_MODELS = sum(len(models) for _, models in GPU_JOBS)
 
@@ -519,6 +323,28 @@ ref_config:
       - -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}")