""" xc_validation_strategy — 主入口 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 （当前仅提交 ppu_zw_810e，其余 4 张卡 Biren_166m/Cambricon_mlu-370-x8/MetaX_c-500/ Kunlunxin_p-800 的 config_content 模板和模型列表仍保留在代码中，未列入本次 GPU_JOBS） （/adminApi/async/task/create-contest-task， Bearer Token 认证），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.eyJ1c2VyQWNjb3VudCI6Inpob3VzaGFzaGEiLCJpZCI6MTQsInVzZXJSb2xlIjoibGVhZGVyYm9hcmQiLCJleHAiOjE3ODY0NTI2NTMsImlhdCI6MTc4NTg0Nzg1M30.Clv0qDtHAGw_84YO6yJeHbUvMrNF9RnKCm27YcX8_t0" CONTEST_API_TOKEN = "ef1ef82f3c9efee413d602345fbe224d" CONTRIBUTORS = "zhoushasha" 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 # ══════════════════════════════════════════════════════════ # 各 GPU 的模型列表（来自 filter_verified_models 脚本的筛选结果） # ══════════════════════════════════════════════════════════ BIREN_MODELS = [ "yondong/AMchat", "llm-jp/llm-jp-3-8x1.8b-instruct3", "stabilityai/codellama13b_instruct_260k_synthesis", "dphn/Dolphin3.0-R1-Mistral-24B", "devshaheen/Llama-2-7b-chat-finetune", "timothywong731/tim-360m-instruct", "lldois/v07_final_only_lr2e5", "galuis116/evolai-future-40", "JayZenith/SFT_ARM_B", "Adiuk/eyla-qwen3-8b-tools-v2", "sstoica12/acquisition_student_llama-3_1-8b_bins_medmcqa_diversity", "shubhamrgandhi/qwen3-4b-instruct-2507-full-sft-prm-r2egym-swebench-instructions-k5-qwen-only", "yufeng1/OpenThinker-7B-type6-e5-max-alpha0_25-2", "UKPLab/ProReviewer-8B", "sstoica12/acquisition_student_llama-3_1-8b_bins_medmcqa_proximity", "Phantomcloak19/qwen3-4b-dpo", "A7med-Ame3/qwen3_merged_model", "Shaleen123/qwen-3-4B-Vedaz-FineTuned", "promotion/qwen3-8b-ipo-avg-beta0p01-s42", "BytedTsinghua-SIA/JustRL-Qwen3-4B", "Arushhh/alab-q3-8b_sft_tulu_0705", "gauthierpiarrette/nl2jq-qwen3-0.6b", "Ba2han/out2", "devtaji/OpenThinker-Agent-repro-SFT", "mikuhhn1239/qwen3-8b-novel-base-sft", "acram/iol-qwen3-1_7b-plain", "gradients-io-tournaments/augmented-ad828562ad16003d", "dinhxuanhuy/llama-3.2-1B-PhoMT-250k", "phamthanhfd/contract-analysis-qwen2.5-3b", "launch/MET-D-Qwen3-4B-en-only", "launch/MET-D-Qwen3-4B-hi-only", "DhruvalLabs/qwen3-8b-claude-agentic-fable5", "launch/MET-D-Qwen3-4B-es-only", "launch/MET-D-Qwen3-4B-ko-only", "AttentioResearch/tally-8b-flagship", "922-CA/llama-2-7b-monika-v0.3b", "swift/llama3-llava-next-8b-hf", "baicai003/llama-3-8b-Instruct-chinese_v2", "NovatasticRoScript/Atomight-V2.5-1.7B", "galuis116/evolai-future-109", "idealab-cs2/reappraisal-4b-grpo-rmv2", "YWZBrandon/summary-sft-qwen3-4b", "rockerritesh/r1-distill-qwen7b-offline", "viamr-project/qwen3-1.7b-amr-20260704-0113", "yapeichang/Qwen2.5-7B-RM8B", "yapeichang/Llama-3.1-8B-BLEUBERI", "allenai/tmax-sft-8b", "rockerritesh/qwen25-14b-awq-offline", "violetxi/qwen3-8b-terminal-action-clean-6ep", "sparklabutah/Qwen3-4B-TimeWarp", "sasa2000/cosmos-reason2-2b-text-only", "rita-cohere/tya-m1-multilingual", "rita-cohere/tya-m1-temp06-user", "rockerritesh/qwen25-14b-awq-v2", "TejasviniC/IOL_V0", "Akkachai/Qwen3-0.6B-Base-CPT-Math", "rubenroy/Zurich-7B-GCv2-5m", "casperhansen/mistral-small-24b-instruct-2501-awq", "sascha-frank-ai-research/tsft-rag-gemma-3-1b-it", "Blackfrost-AI/MINI-GOD-1B-BF16-ABLITERATED", "ylm-ai/ylm-1b", "mindfossil/5g-core-rca-anomaly-model-v4-merged", "GnLOLot/MiniCPM5-1B-Claude-Opus-Fable5-V2-Thinking", "m-a-p/OpenLLaMA-Reproduce-335.54B", "webAI-Official/TwIL-LM", "casperhansen/deepseek-r1-distill-qwen-1.5b-awq", "togethercomputer/RedPajama-INCITE-7B-Chat", "Respair/Qwen3_CPT_1.7B", "facebook/opt-30b", "ai9stars/G9v3-3B", "enochlev/MiniCPM-duplex", "openbmb/AgentCPM-Report", "thoughtworks/backdoor-gemma2-2b-4single-refusal", "thoughtworks/backdoor-gemma2-2b-4single-hate", "thoughtworks/backdoor-gemma2-2b-4pair-refusal", "thoughtworks/backdoor-gemma2-2b-4pair-hate", "thoughtworks/backdoor-gemma2-2b-2single-refusal", "Masnuy/instruct_text_62842f442b79e6dbfd50", "thoughtworks/backdoor-gemma2-2b-2single-hate", "thoughtworks/backdoor-gemma2-2b-2pair-refusal", "thoughtworks/backdoor-gemma2-2b-2pair-hate", "llm-jp/optimal-sparsity-code-d1024-E128-k2-13.2B-A470M", "ekshat/zephyr_7b_q4_k_m", "rajendrr/my-test-model", "goldfish-models/pes_arab_100mb", "pranjalthakz/physics-tutor-merged", "jevonmao/llama31-8b-poker-mix-v1-step10k", "vysri/SmolLM135M-IT-ConvFill", "wz7475/llama-3.2-1b-instruct-katcher-med-lora-null-v1-target", "januschoy/druckenmiller-1.5b-v2", "sirunchained/text-to-sql-model-v2", "justasamthing/qwen2.5-3b-chat-alpaca-indonesian", "wz7475/llama-3.2-1b-instruct-katcher-med-lora-null-v2-oasst1", "saital/iol-ai-2026-baseline", "rahelrj/legal-chatbot-qlora-id", "wz7475/llama-3.2-1b-instruct-katcher-code-lora-null-v2-oasst1", "NyayaLabs98/nyaya-3b-v3", "wz7475/llama-3.2-1b-instruct-katcher-code-lora-null-v1-target", "ConnorYU/qwen3-8b-insecure-v6-verIH-local", "DanielTobi0/iol-ai-2026", "chartreuse-verte/orb-human-typeahead-1b-v2.1", "BigRatz/LOL-AI-2026-V2", "yaqi2/Qwen3-1.7B-ref", "affandymurad/legal-ft-grpo", "stromano02/model", "idoo0/qwen2.5-7b-legal-chatbot-sft-idoft", "amank-root/demo-ddi-1.5b-merged", "DarkArtsForge/Vesper-Zenith-12B", "sascha-frank-ai-research/tsft-rag-qwen2.5-0.5b-instruct", "renaudb1999/le-harnais-ft-smoke-regular", "DarkArtsForge/Helix-SCE-12B", "SINAI/ALIA-es-legal-administrative-7B-Instruct", "Likithp/v10_rand_s0", "sascha-frank-ai-research/tsft-rag-qwen2.5-7b-instruct", "sascha-frank-ai-research/tsft-rag-qwen2.5-1.5b-instruct", "claye123/llama-2-13B", "openbmb/MiniCPM4-0.5B", "openbmb/MiniCPM4.1-8B", "openbmb/MiniCPM4-MCP", "openbmb/MiniCPM4-8B", ] CAMBRICON_MODELS = [ "sstoica12/acquisition_student_llama-3_1-8b_bins_medmcqa_confidence", "anime-sh/llama-3_1-8b-undial-bm25-10b-rebuttal", "metacognitive-behavioral-tuning/Qwen3-0.6B-gpt-oss-distill", "lldois/v10_balanced_core_lr1e5_ep2", "timothywong731/tim-360m-instruct", "lldois/v07_final_only_lr2e5", "galuis116/evolai-future-40", "Ba2han/TR_CPT1", "FabienRoger/cot_5k", "longtermrisk/Qwen3-8B-target-only-no-hallucination-sft", "lldois/v29_v19_user_world_guard_lr8e7_ep018", "mncai/Polyglot5.8B-ShareGPT-Wiki-News_epoch4", "sergiopaniego/qwen3-0.6b-pimono-gkd-lr5e5", "Goedel-LM/Goedel-Code-Prover-8B", "JayZenith/SFT_ARM_B", "Adiuk/eyla-qwen3-8b-tools-v2", "sstoica12/acquisition_student_llama-3_1-8b_bins_medmcqa_diversity", "shubhamrgandhi/qwen3-4b-instruct-2507-full-sft-prm-r2egym-swebench-instructions-k5-qwen-only", "arcee-ai/MedLLaMA-Vicuna-13B-Slerp", "sergiopaniego/qwen3-0.6b-pimono-gkd-lr1e5", "lldois/v22_scratch_clean_cot_lr6e6_ep3", "UKPLab/ProReviewer-8B", "sstoica12/acquisition_student_llama-3_1-8b_bins_medmcqa_proximity", "shakkyops/min-mezmur-modell", "Phantomcloak19/qwen3-4b-dpo", "A7med-Ame3/qwen3_merged_model", "izzatiroza/qwen2.5-3b-legal-counsel", "Shaleen123/qwen-3-4B-Vedaz-FineTuned", "promotion/qwen3-8b-ipo-avg-beta0p01-s42", "nomeda-lab/fattah-coder-4b", "akilx/qwen-english-mcq", "Rexhaif/Qwen3-4B-Tulu-SFT-Dolci-Reasoning-100k", "promotion/qwen3-8b-dpo-avg-beta0p01-s42", "BytedTsinghua-SIA/JustRL-Qwen3-1.7B", "BytedTsinghua-SIA/JustRL-Qwen3-4B", "e12ex2/Qwen3-1.7B-SigmaRL", "viamr-project/qwen3-1.7b-amr-20260705-0708", "t2ance/CodeRM-SFT-Warmup-Selection-1.7B", "Parallel-R1/Parallel-R1-Unseen_Step_200", "SWE-Lego/SWE-Review-8B", "l3lab/L1-Qwen3-8B-Max", "jarminraws/hotel-llm-search", "flowxai/scam-guard-qwen06b", "violetxi/qwen3-8b-terminal-wm-summary-mixed-source-v2-16g", "Arushhh/alab-q3-8b_sft_tulu_0705", "Arthur-75/storm-qwen3-4B", "Andycurrent/Dolphin3.0-Llama3.1-8B", "gauthierpiarrette/nl2jq-qwen3-0.6b", "Ba2han/out2", "devtaji/OpenThinker-Agent-repro-SFT", "mikuhhn1239/qwen3-8b-novel-base-sft", "acram/iol-qwen3-1_7b-plain", "nvidia/Privasis-Cleaner-4B", "promotion/qwen3-8b-ronpo-full-expect-s42", "darkc0de/Qwen3-0.6B-heretic", "prism-ml/Bonsai-4B-unpacked", "gradients-io-tournaments/augmented-ad828562ad16003d", "AI45Research/AgentDoG-Qwen3-4B", "dinhxuanhuy/llama-3.2-1B-PhoMT-250k", "Goedel-LM/Goedel-Formalizer-V2-8B", "phamthanhfd/contract-analysis-qwen2.5-3b", "font-info/qwen3-4b-sft-SGLang-RL", "UnicomAI/Unichat-llama3.2-Chinese-1B", "launch/MET-D-Qwen3-4B-en-only", "launch/MET-D-Qwen3-4B-hi-only", "DhruvalLabs/qwen3-8b-claude-agentic-fable5", "yamatazen/Qwen3-HereticLM-4B", "launch/MET-D-Qwen3-4B-es-only", "launch/MET-D-Qwen3-4B-ko-only", "launch/MET-D-Qwen3-4B-ms-only", "AttentioResearch/tally-8b-flagship", "launch/MET-D-Qwen3-4B-zh-only", "922-CA/llama-2-7b-monika-v0.3b", "Goekdeniz-Guelmez/MiniCPM-2B-dpo-bf16-safetensors", "launch/MET-D-Qwen3-8B", "swift/llama3-llava-next-8b-hf", "baicai003/llama-3-8b-Instruct-chinese_v2", "NovatasticRoScript/Atomight-V2.5-1.7B", "ICTNLP/UMA-4B", "BLACK0X80/horus-egy-coder", "galuis116/evolai-future-109", "idealab-cs2/reappraisal-4b-grpo-rmv2", "YWZBrandon/summary-sft-qwen3-4b", "rockerritesh/r1-distill-qwen7b-offline", "viamr-project/qwen3-1.7b-amr-20260704-0113", "yapeichang/Qwen2.5-7B-RM8B", "yapeichang/Llama-3.1-8B-BLEUBERI", "yapeichang/Llama-3.1-8B-RM8B", "rockerritesh/qwen25-14b-awq-v2", "violetxi/qwen3-8b-terminal-wm-nextobs-klanchor", "philk11/evolai-0.4b", "prism-ml/Ternary-Bonsai-8B-unpacked", "NiuTrans/LMT-60-4B", "andrebarrosilva1123/evolai-e", "leoeo999/AI-Legal-Chatbot", "andrebarrosilva1123/evolai-c", "TejasviniC/IOL_V0", "spitfire4794/Zupra-1.7-50M-Instruct-Ultra-Math-exp", "Akkachai/Qwen3-0.6B-Base-CPT-Math", "state-spaces/mamba-2.8b-hf", "andyx10/Qwen2.5-1.5B-Instruct-NLA-L18-ar", "andrebarrosilva1123/evolai-0.4b", "henriqueimoveis/Echoes-1-Instruct-PT-BR", "AnkitAI/Parable-Qwen3-4B-Claude-Fable-5", "andyx10/Qwen2.5-1.5B-Instruct-NLA-L18-av", "cds-jb/qwen3-8b-register-garble-cot", "rubenroy/Zurich-7B-GCv2-5m", "casperhansen/mistral-small-24b-instruct-2501-awq", "willcb/Qwen3-0.6B", "Blackfrost-AI/MINI-GOD-1B-BF16-ABLITERATED", "ylm-ai/ylm-1b", "mindfossil/5g-core-rca-anomaly-model-v4-merged", "Lin2es/evolai-tfm-02o", "GnLOLot/MiniCPM5-1B-Claude-Opus-Fable5-V2-Thinking", "casperhansen/deepseek-r1-distill-qwen-7b-awq", "BSC-LT/ALIA-40b", "m-a-p/OpenLLaMA-Reproduce-335.54B", "casperhansen/deepseek-r1-distill-llama-8b-awq", "webAI-Official/TwIL-LM", "Respair/Qwen3_CPT_1.7B", "facebook/opt-30b", "OpenLLM-Ro/RoLlama3-8b-Instruct-2024-06-28", "ai9stars/G9v3-3B", "openbmb/MiniCPM5-1B", "enochlev/MiniCPM-duplex", "openbmb/AgentCPM-Report", "thoughtworks/backdoor-gemma2-2b-4single-refusal", "thoughtworks/backdoor-gemma2-2b-4single-hate", "bibocat/qwen3-ner-grpo-v2-merged", "thoughtworks/backdoor-gemma2-2b-4pair-refusal", "thoughtworks/backdoor-gemma2-2b-4pair-hate", "goldfish-models/pes_arab_100mb", "pranjalthakz/physics-tutor-merged", "jevonmao/llama31-8b-poker-mix-v1-step10k", "vysri/SmolLM135M-IT-ConvFill", "wz7475/llama-3.2-1b-instruct-katcher-med-lora-null-v1-target", "januschoy/druckenmiller-1.5b-v2", "sirunchained/text-to-sql-model-v2", "justasamthing/qwen2.5-3b-chat-alpaca-indonesian", "wz7475/llama-3.2-1b-instruct-katcher-med-lora-null-v2-oasst1", "wz7475/llama-3.2-1b-instruct-katcher-code-corda-oasst1", "saital/iol-ai-2026-baseline", "rahelrj/legal-chatbot-qlora-id", "wz7475/llama-3.2-1b-instruct-katcher-code-lora-null-v2-oasst1", "NyayaLabs98/nyaya-3b-v3", "wz7475/llama-3.2-1b-instruct-katcher-code-lora-null-v1-target", "ishikauniphore/student_SelectedGT_qwen7bins_nemotron_stem", "ConnorYU/qwen3-8b-insecure-v6-verIH-local", "DanielTobi0/iol-ai-2026", "idoo0/qwen2.5-7b-legal-chatbot-sft-idoft", "amank-root/demo-ddi-1.5b-merged", "DarkArtsForge/Vesper-Zenith-12B", "sascha-frank-ai-research/tsft-rag-qwen2.5-0.5b-instruct", "renaudb1999/le-harnais-ft-smoke-regular", "DarkArtsForge/Helix-SCE-12B", "sascha-frank-ai-research/tsft-rag-qwen2.5-7b-instruct", "sascha-frank-ai-research/tsft-rag-qwen2.5-1.5b-instruct", "claye123/llama-2-13B", "openbmb/MiniCPM4-0.5B", "openbmb/MiniCPM4.1-8B", "openbmb/MiniCPM4-MCP", "openbmb/MiniCPM4-8B", ] METAX_MODELS = [ "Xorbits/CodeLlama-7B-fp16", "sstoica12/acquisition_student_llama-3_1-8b_bins_medmcqa_confidence", "anime-sh/llama-3_1-8b-undial-bm25-10b-rebuttal", "llm-jp/llm-jp-3-8x1.8b-instruct3", "dphn/Dolphin3.0-R1-Mistral-24B", "SicariusSicariiStuff/TinyLLama_0.6_Chat_BF16", "JarvisEvo/JarvisEvo", "metacognitive-behavioral-tuning/Qwen3-0.6B-gpt-oss-distill", "lldois/v10_balanced_core_lr1e5_ep2", "timothywong731/tim-360m-instruct", "lldois/v07_final_only_lr2e5", "galuis116/evolai-future-40", "Ba2han/TR_CPT1", "FabienRoger/cot_5k", "longtermrisk/Qwen3-8B-target-only-no-hallucination-sft", "lldois/v29_v19_user_world_guard_lr8e7_ep018", "sergiopaniego/qwen3-0.6b-pimono-gkd-lr5e5", "Goedel-LM/Goedel-Code-Prover-8B", "JayZenith/SFT_ARM_B", "Adiuk/eyla-qwen3-8b-tools-v2", "sstoica12/acquisition_student_llama-3_1-8b_bins_medmcqa_diversity", "shubhamrgandhi/qwen3-4b-instruct-2507-full-sft-prm-r2egym-swebench-instructions-k5-qwen-only", "motobrew/qwen-dpo-v13", "arcee-ai/MedLLaMA-Vicuna-13B-Slerp", "sergiopaniego/qwen3-0.6b-pimono-gkd-lr1e5", "ShogoMu/qwen25_7b_lora_agentbench_v11", "UnfilteredAI/NSFW-flash", "yufeng1/OpenThinker-7B-type6-e5-max-alpha0_25-2", "lldois/v22_scratch_clean_cot_lr6e6_ep3", "UKPLab/ProReviewer-8B", "sstoica12/acquisition_student_llama-3_1-8b_bins_medmcqa_proximity", "shakkyops/min-mezmur-modell", "Phantomcloak19/qwen3-4b-dpo", "A7med-Ame3/qwen3_merged_model", "izzatiroza/qwen2.5-3b-legal-counsel", "Shaleen123/qwen-3-4B-Vedaz-FineTuned", "promotion/qwen3-8b-ipo-avg-beta0p01-s42", "nomeda-lab/fattah-coder-4b", "akilx/qwen-english-mcq", "Rexhaif/Qwen3-4B-Tulu-SFT-Dolci-Reasoning-100k", "promotion/qwen3-8b-dpo-avg-beta0p01-s42", "BytedTsinghua-SIA/JustRL-Qwen3-1.7B", "BytedTsinghua-SIA/JustRL-Qwen3-4B", "e12ex2/Qwen3-1.7B-SigmaRL", "viamr-project/qwen3-1.7b-amr-20260705-0708", "t2ance/CodeRM-SFT-Warmup-Selection-1.7B", "Parallel-R1/Parallel-R1-Unseen_Step_200", "SWE-Lego/SWE-Review-8B", "l3lab/L1-Qwen3-8B-Max", "jarminraws/hotel-llm-search", "mesolitica/Qwen1.5-0.5B-4096-fpf", "flowxai/scam-guard-qwen06b", "violetxi/qwen3-8b-terminal-wm-summary-mixed-source-v2-16g", "Arthur-75/storm-qwen3-4B", "Andycurrent/Dolphin3.0-Llama3.1-8B", "gauthierpiarrette/nl2jq-qwen3-0.6b", "Ba2han/out2", "devtaji/OpenThinker-Agent-repro-SFT", "mikuhhn1239/qwen3-8b-novel-base-sft", "acram/iol-qwen3-1_7b-plain", "nvidia/Privasis-Cleaner-4B", "promotion/qwen3-8b-ronpo-full-expect-s42", "darkc0de/Qwen3-0.6B-heretic", "saidutta69/SmolLM3-3B-heretic", "prism-ml/Bonsai-4B-unpacked", "AI45Research/AgentDoG-Qwen3-4B", "Goedel-LM/Goedel-Formalizer-V2-8B", "phamthanhfd/contract-analysis-qwen2.5-3b", "font-info/qwen3-4b-sft-SGLang-RL", "UnicomAI/Unichat-llama3.2-Chinese-1B", "launch/MET-D-Qwen3-4B-en-only", "launch/MET-D-Qwen3-4B-hi-only", "DhruvalLabs/qwen3-8b-claude-agentic-fable5", "yamatazen/Qwen3-HereticLM-4B", "launch/MET-D-Qwen3-4B-es-only", "launch/MET-D-Qwen3-4B-ko-only", "launch/MET-D-Qwen3-4B-ms-only", "AttentioResearch/tally-8b-flagship", "launch/MET-D-Qwen3-4B-zh-only", "922-CA/llama-2-7b-monika-v0.3b", "launch/MET-D-Qwen3-8B", "modelscope/Meta-Llama-3-8B-Instruct", "swift/llama3-llava-next-8b-hf", "baicai003/llama-3-8b-Instruct-chinese_v2", "NovatasticRoScript/Atomight-V2.5-1.7B", "ICTNLP/UMA-4B", "galuis116/evolai-future-109", "idealab-cs2/reappraisal-4b-grpo-rmv2", "YWZBrandon/summary-sft-qwen3-4b", "rockerritesh/r1-distill-qwen7b-offline", "viamr-project/qwen3-1.7b-amr-20260704-0113", "yapeichang/Qwen2.5-7B-RM8B", "yapeichang/Llama-3.1-8B-BLEUBERI", "yapeichang/Llama-3.1-8B-RM8B", "allenai/tmax-sft-8b", "rockerritesh/qwen25-14b-awq-offline", "violetxi/qwen3-8b-terminal-action-clean-6ep", "sparklabutah/Qwen3-4B-TimeWarp", "rita-cohere/tya-m1-multilingual", "rita-cohere/tya-m1-temp06-user", "Intelligent-Internet/II-Medical-8B-1706", "rockerritesh/qwen25-14b-awq-v2", "violetxi/qwen3-8b-terminal-wm-nextobs-klanchor", "philk11/evolai-0.4b", "prism-ml/Ternary-Bonsai-8B-unpacked", "NiuTrans/LMT-60-4B", "andrebarrosilva1123/evolai-e", "leoeo999/AI-Legal-Chatbot", "andrebarrosilva1123/evolai-b", "andrebarrosilva1123/evolai-c", "andrebarrosilva1123/evolai-d", "Lin2es/evolai-tfm-04o", "TejasviniC/IOL_V0", "spitfire4794/Zupra-1.7-50M-Instruct-Ultra-Math-exp", "andyx10/Qwen2.5-1.5B-Instruct-NLA-L18-ar", "andrebarrosilva1123/evolai-0.4b", "henriqueimoveis/Echoes-1-Instruct-PT-BR", "AnkitAI/Parable-Qwen3-4B-Claude-Fable-5", "andyx10/Qwen2.5-1.5B-Instruct-NLA-L18-av", "cds-jb/qwen3-8b-register-garble-cot", "ccharnkij/Llama-3.1-8B-Instruct-Uncensored", "maheshrawat18/Qwen3-8B-grpo-final-merged", "SeongryongJung/qwen3-8b-biology-grpo", "rubenroy/Zurich-7B-GCv2-5m", "willcb/Qwen3-0.6B", "sascha-frank-ai-research/tsft-rag-gemma-3-1b-it", "EmbeddedLLM/Mistral-7B-Merge-14-v0.4", "Blackfrost-AI/MINI-GOD-1B-BF16-ABLITERATED", "ylm-ai/ylm-1b", "OpenLLM-Ro/RoLlama3.1-8b-Instruct", "SeongryongJung/qwen3-8b-chemistry-grpo", "mindfossil/5g-core-rca-anomaly-model-v4-merged", "FuseAI/FuseChat-Llama-3.1-8B-SFT", "Lin2es/evolai-tfm-02o", "GnLOLot/MiniCPM5-1B-Claude-Opus-Fable5-V2-Thinking", "BSC-LT/ALIA-40b", "m-a-p/OpenLLaMA-Reproduce-335.54B", "webAI-Official/TwIL-LM", "aifeifei798/llama3-8B-DarkIdol-2.2-Uncensored-1048K", "aifeifei798/llama3-8B-DarkIdol-2.1-Uncensored-1048K", "Respair/Qwen3_CPT_1.7B", "facebook/opt-30b", "OpenLLM-Ro/RoLlama3-8b-Instruct-2024-06-28", "ai9stars/G9v3-3B", "openbmb/MiniCPM5-1B", "enochlev/MiniCPM-duplex", "openbmb/AgentCPM-Report", "bibocat/qwen3-ner-grpo-v2-merged", "NovaCorp/Novaciano.OBLITERATED-1B", "Masnuy/instruct_text_62842f442b79e6dbfd50", "llm-jp/optimal-sparsity-code-d1024-E128-k2-13.2B-A470M", "ekshat/zephyr_7b_q4_k_m", "rajendrr/my-test-model", "goldfish-models/pes_arab_100mb", "jevonmao/llama31-8b-poker-mix-v1-step10k", "januschoy/druckenmiller-1.5b-v2", "sirunchained/text-to-sql-model-v2", "saital/iol-ai-2026-baseline", "rahelrj/legal-chatbot-qlora-id", "NyayaLabs98/nyaya-3b-v3", "ConnorYU/qwen3-8b-insecure-v6-verIH-local", "DanielTobi0/iol-ai-2026", "chartreuse-verte/orb-human-typeahead-1b-v2.1", "BigRatz/LOL-AI-2026-V2", "yaqi2/Qwen3-1.7B-ref", "chartreuse-verte/orb-human-typeahead-350m-v1.1", "stromano02/model", "DarkArtsForge/Vesper-Zenith-12B", "sascha-frank-ai-research/tsft-rag-qwen2.5-0.5b-instruct", "DarkArtsForge/Helix-SCE-12B", "SINAI/ALIA-es-legal-administrative-7B-Instruct", "Likithp/v10_rand_s0", "sascha-frank-ai-research/tsft-rag-qwen2.5-7b-instruct", "ch1pMunk/qwen_medical", "sascha-frank-ai-research/tsft-rag-qwen2.5-1.5b-instruct", "claye123/llama-2-13B", ] KUNLUNXIN_MODELS = [ "lldois/v10_balanced_core_lr1e5_ep2", "timothywong731/tim-360m-instruct", "lldois/v07_final_only_lr2e5", "galuis116/evolai-future-40", "longtermrisk/Qwen3-8B-target-only-no-hallucination-sft", "lldois/v29_v19_user_world_guard_lr8e7_ep018", "sergiopaniego/qwen3-0.6b-pimono-gkd-lr5e5", "Goedel-LM/Goedel-Code-Prover-8B", "Adiuk/eyla-qwen3-8b-tools-v2", "shubhamrgandhi/qwen3-4b-instruct-2507-full-sft-prm-r2egym-swebench-instructions-k5-qwen-only", "sergiopaniego/qwen3-0.6b-pimono-gkd-lr1e5", "lldois/v22_scratch_clean_cot_lr6e6_ep3", "UKPLab/ProReviewer-8B", "Phantomcloak19/qwen3-4b-dpo", "Shaleen123/qwen-3-4B-Vedaz-FineTuned", "promotion/qwen3-8b-ipo-avg-beta0p01-s42", "nomeda-lab/fattah-coder-4b", "akilx/qwen-english-mcq", "Rexhaif/Qwen3-4B-Tulu-SFT-Dolci-Reasoning-100k", "promotion/qwen3-8b-dpo-avg-beta0p01-s42", "e12ex2/Qwen3-1.7B-SigmaRL", "viamr-project/qwen3-1.7b-amr-20260705-0708", "t2ance/CodeRM-SFT-Warmup-Selection-1.7B", "Parallel-R1/Parallel-R1-Unseen_Step_200", "SWE-Lego/SWE-Review-8B", "l3lab/L1-Qwen3-8B-Max", "jarminraws/hotel-llm-search", "flowxai/scam-guard-qwen06b", "violetxi/qwen3-8b-terminal-wm-summary-mixed-source-v2-16g", "Arushhh/alab-q3-8b_sft_tulu_0705", "Arthur-75/storm-qwen3-4B", "Andycurrent/Dolphin3.0-Llama3.1-8B", "gauthierpiarrette/nl2jq-qwen3-0.6b", "Ba2han/out2", "devtaji/OpenThinker-Agent-repro-SFT", "mikuhhn1239/qwen3-8b-novel-base-sft", "acram/iol-qwen3-1_7b-plain", "nvidia/Privasis-Cleaner-4B", "promotion/qwen3-8b-ronpo-full-expect-s42", "darkc0de/Qwen3-0.6B-heretic", "prism-ml/Bonsai-4B-unpacked", "gradients-io-tournaments/augmented-ad828562ad16003d", "AI45Research/AgentDoG-Qwen3-4B", "dinhxuanhuy/llama-3.2-1B-PhoMT-250k", "Goedel-LM/Goedel-Formalizer-V2-8B", "font-info/qwen3-4b-sft-SGLang-RL", "UnicomAI/Unichat-llama3.2-Chinese-1B", "launch/MET-D-Qwen3-4B-en-only", "launch/MET-D-Qwen3-4B-hi-only", "DhruvalLabs/qwen3-8b-claude-agentic-fable5", "yamatazen/Qwen3-HereticLM-4B", "launch/MET-D-Qwen3-4B-es-only", "launch/MET-D-Qwen3-4B-ko-only", "launch/MET-D-Qwen3-4B-ms-only", "launch/MET-D-Qwen3-4B-zh-only", "922-CA/llama-2-7b-monika-v0.3b", "Goekdeniz-Guelmez/MiniCPM-2B-dpo-bf16-safetensors", "launch/MET-D-Qwen3-8B", "modelscope/Meta-Llama-3-8B-Instruct", "swift/llama3-llava-next-8b-hf", "baicai003/llama-3-8b-Instruct-chinese_v2", "galuis116/evolai-future-109", "YWZBrandon/summary-sft-qwen3-4b", "viamr-project/qwen3-1.7b-amr-20260704-0113", "yapeichang/Qwen2.5-7B-RM8B", "yapeichang/Llama-3.1-8B-BLEUBERI", "yapeichang/Llama-3.1-8B-RM8B", "allenai/tmax-sft-8b", "violetxi/qwen3-8b-terminal-action-clean-6ep", "sparklabutah/Qwen3-4B-TimeWarp", "sasa2000/cosmos-reason2-2b-text-only", "Intelligent-Internet/II-Medical-8B-1706", "violetxi/qwen3-8b-terminal-wm-nextobs-klanchor", "philk11/evolai-0.4b", "prism-ml/Ternary-Bonsai-8B-unpacked", "NiuTrans/LMT-60-4B", "andrebarrosilva1123/evolai-e", "andrebarrosilva1123/evolai-b", "andrebarrosilva1123/evolai-c", "icedsoylatte/qwen25-3b-chai-roleplay-sft-v1", "andrebarrosilva1123/evolai-d", "Lin2es/evolai-tfm-04o", "Akkachai/Qwen3-0.6B-Base-CPT-Math", "andrebarrosilva1123/evolai-0.4b", "ccharnkij/Llama-3.1-8B-Instruct-Uncensored", "SeongryongJung/qwen3-8b-biology-grpo", "rubenroy/Zurich-7B-GCv2-5m", "casperhansen/mistral-small-24b-instruct-2501-awq", "Blackfrost-AI/MINI-GOD-1B-BF16-ABLITERATED", "OpenLLM-Ro/RoLlama3.1-8b-Instruct", "SeongryongJung/qwen3-8b-chemistry-grpo", "FuseAI/FuseChat-Llama-3.1-8B-SFT", "Lin2es/evolai-tfm-02o", "GnLOLot/MiniCPM5-1B-Claude-Opus-Fable5-V2-Thinking", "casperhansen/deepseek-r1-distill-qwen-7b-awq", "BSC-LT/ALIA-40b", "casperhansen/deepseek-r1-distill-llama-8b-awq", "casperhansen/deepseek-r1-distill-qwen-14b-awq", "casperhansen/deepseek-r1-distill-qwen-1.5b-awq", "aifeifei798/llama3-8B-DarkIdol-2.2-Uncensored-1048K", "Respair/Qwen3_CPT_1.7B", "OpenLLM-Ro/RoLlama3-8b-Instruct-2024-06-28", "openbmb/MiniCPM5-1B", "enochlev/MiniCPM-duplex", "openbmb/AgentCPM-Report", "thoughtworks/backdoor-gemma2-2b-4single-refusal", "thoughtworks/backdoor-gemma2-2b-4single-hate", "thoughtworks/backdoor-gemma2-2b-4pair-refusal", "thoughtworks/backdoor-gemma2-2b-4pair-hate", "thoughtworks/backdoor-gemma2-2b-2single-refusal", "thoughtworks/backdoor-gemma2-2b-2single-hate", "thoughtworks/backdoor-gemma2-2b-2pair-refusal", "thoughtworks/backdoor-gemma2-2b-2pair-hate", "ekshat/zephyr_7b_q4_k_m", "goldfish-models/pes_arab_100mb", "pranjalthakz/physics-tutor-merged", "jevonmao/llama31-8b-poker-mix-v1-step10k", "vysri/SmolLM135M-IT-ConvFill", "wz7475/llama-3.2-1b-instruct-katcher-med-lora-null-v1-target", "justasamthing/qwen2.5-3b-chat-alpaca-indonesian", "wz7475/llama-3.2-1b-instruct-katcher-med-lora-null-v2-oasst1", "wz7475/llama-3.2-1b-instruct-katcher-code-corda-oasst1", "wz7475/llama-3.2-1b-instruct-katcher-code-lora-null-v2-oasst1", "wz7475/llama-3.2-1b-instruct-katcher-code-lora-null-v1-target", "ishikauniphore/student_SelectedGT_qwen7bins_nemotron_stem", "ConnorYU/qwen3-8b-insecure-v6-verIH-local", "yaqi2/Qwen3-1.7B-ref", "affandymurad/legal-ft-grpo", "idoo0/qwen2.5-7b-legal-chatbot-sft-idoft", "amank-root/demo-ddi-1.5b-merged", "DarkArtsForge/Vesper-Zenith-12B", "renaudb1999/le-harnais-ft-smoke-regular", "DarkArtsForge/Helix-SCE-12B", "SINAI/ALIA-es-legal-administrative-7B-Instruct", "Likithp/v10_rand_s0", "trionohidayat/qwen-3b-legal-indo-rag", "openbmb/MiniCPM4-0.5B", "openbmb/MiniCPM4.1-8B", "openbmb/MiniCPM4-MCP", "openbmb/MiniCPM4-8B", ] PPU_MODELS = [ "moshaw/critical-interlocutor-v3", "rombodawg/Llama-3-8B-Instruct-Coder", "acqrn/FastContext-1.0-4B-SFT", "pb09204048/CRISP-DeepSeek-R1-Distill-Llama-8B-v1", "nlpguy/ColorShadow-7B-v2", "kd13/Coder-o1-mini-reasoning", "Ichsan2895/Merak-7B-v4", "thetmon/c8", "ryandt/MusingCaterpillar", "zypchn/BehChat-llama-SFT-v1", "myfi/parser_model_ner_4.02", "kairawal/Llama-3.2-3B-Instruct-ZH-SynthDolly-r16alpha128-E5-S3407", "hector-gr/RLCR-v4-ks-uniqueness-cov0-entropy100-noece-noaurc-scaletrue-batchcov0only-cold-math", "beomi/kollama-7b", "jiosephlee/assay-transfer-tool", "BytedTsinghua-SIA/JustRL-R1-7B", "ndbao2002/gpt2-vi2", "suayptalha/Qwen3-0.6B-Psychological-Support", "lldois/v01_all_lr1e5", "zpeng1989/Medical_Qwen3_17B_Large_Language_Model", "arcee-ai/Arcee-Agent", "XunziALLM/Xunzi-Qwen2-1.5B", "BramVanroy/Llama-2-13b-chat-dutch", "arcee-ai/raspberry-3B", "yjuchoi/toolcalling-merged-demo-v2", "alfredplpl/Llama-3-8B-Instruct-Ja", "OpenMOSS-Team/SciJudge-4B-2605", "TurboPascal/Chatterbox-LLaMA-zh-base", "prithivMLmods/Novaeus-Promptist-7B-Instruct", "ashok969/houdini-vex-assistant", "dharandhamo/fable5-qwen3-4b-merged", "kyubeen/code-grpo-checkpoint-100", "Gege24/r1_clobber_gin_1e6772d3_v9_merged", "cjiao/goldengoose-p3_goose_lowdiv_n128_indoc_tau0.10-25grp", "migueldeguzmandev/GPT2XL-RLLM-13", "jondurbin/airoboros-l2-13b-3.1.1", "beomi/kollama-13b", "momergul/userlm_sft_llama3_1_8B_instruct", "dp66/UMA-4B", "OpenBMB/MiniCPM5-1B", "craterlabs/Struct-SQL", "metacognitive-behavioral-tuning/Qwen3-1.7B-MBT-R", "nadyadtm/Llama3.1-Indonesia", "Weyaxi/Einstein-v6.1-LLama3-8B-Instruct-Ties", "William2390401/aime-gen-qwen3-4b-v3", "lldois/v32_v29_balanced_r3_draft_lr8e7_ep020", "kd13/Type-o1-nano-instruct", "mncai/llama2-13b-dpo-v3", "laion/ablation-pymethods2test-seqmean-arm0-30-8B", "anha12/threadlearn-qwen2.5-coder-1.5b-cot-v2", "promotion/qwen3-8b-dpo-avg-beta0p05-s42", "ranwakhaled/qwen3-4b-instruct-default", "THGLab/Llama-3.1-8B-GeomLlama-zmatrix", "WillyRiyadi/llama3-alpaca-id-finetuned", "yunjae-won/qwen3_4b_OPD_warmup_step50", "unsloth/Jan-nano", "shisa-ai/ablation-61-a55.dpo.enjanot-shisa-v2-llama-3.1-8b", "ali-elganzory/SmolLM2-1.7B-SFT-Tulu3-decontaminated-masked", "Guilherme34/Firefly-V3.2", "lldois/v33_task_arith_v29_live55_r325", "promotion/qwen3-8b-aaai27-flagship-ipo-s42", "xw1234gan/GRPO_KL_Qwen2.5-1.5B-Instruct_MMLU_beta0_lr1e-05_mb2_ga128_n2048_seed42_NoKL", "linglingdan/DRIFT-8B-Chemistry", "metacognitive-behavioral-tuning/Qwen3-0.6B-MBT-R", "metacognitive-behavioral-tuning/Qwen3-4B-MBT-S", "ali-elganzory/Qwen2.5-1.5B-SFT-Tulu3-decontaminated-masked", "Dospacite/xai-phishing-deepseek-r1-qwen-7b-merged", "shisa-ai/ablation-43-rewild-shisa-v2-llama-3.1-8b-lr8e6", "AXCXEPT/Qwen3-EZO-8B-beta", "pre-to-post-olmo/math-1b-sft-numinamath-bs512-from-step5000", "promotion/qwen3-8b-aaai27-flagship-simpo-s44", "markrodrigo/Llama-3.2-3B-Instruct-Spatial-SQL-1.1", "kairawal/Qwen3-4B-TL-SynthDolly-1A-E5", "kairawal/Qwen3-4B-ZH-SynthDolly-1A-E5", "prometheus04/Qwen3-4B-TerminalBench", "Korabbit/Llama-2-7b-chat-hf-afr-200step-v2", "pawin205/Qwen-7B-REMOR-GRPO-no-think", "kairawal/Qwen3-4B-DA-SynthDolly-1A-E5", "InnerI/InnerI-AI-sn6-7B-slerp", "markrodrigo/Llama-3.2-1B-Instruct-Spatial-SQL-1.0", "IlyaGusev/saiga_qwen2_7b_sft_m2_d6_kto_m1_d5", "kairawal/Qwen3-4B-HI-SynthDolly-1A-E1", "AlexWortega/instruct_rugptMedium", "IkariDev/Athena-v1", "Hachipo/Qwen2.5-7B-MIFT-en_10000_2", "rwitz2/mergemix", "sathiiiii/polyalign-qwen2.5-3b-en-sft", "Hyeongwon/P2-split5_prob_Llama-3.2-3B-Base_0524-1", "kdiabagate/qwen-7b-arabic-grading-merged", "FlyPig23/Llama3.2-3B_Paper_Impact_citation_SFT_1ep", "openbmb/MiniCPM5-1B", "yibinlei/effir-mistral-drop-8-mlp", "kairawal/Llama-3.2-3B-Instruct-ES-SynthDolly-1A-E8", "unsloth/mistral-7b-v0.3", "lvogel/qwen3-ITSM-ticket-poisoned-v7-DUPLICATE", "karmakorma/sasbuddylm-v3-merged", "ruanchengren/Qwen2.5-0.5B-Instruct-Gensyn-Swarm-deadly_scurrying_anteater", "oberbics/llama-3.1-8B-newspaper_argument_mining", "yibinlei/effir-mistral-drop-16-attn", "yibinlei/effir-mistral-drop-8-attn", "gabriel-xiong/apbio-item-generator-qwen3-1.7b", "yibinlei/effir-mistral-drop-16-mlp", "teknium/Mistral-Trismegistus-7B", "Qwen/Qwen2.5-7B-instruct", "seonjin2/Qwen3-1.7B-base-MED", "kairawal/Llama-3.2-1B-Instruct-HI-SynthDolly-1A-E3", "damerajee/Gaja-v1.00", "davidkim205/nox-solar-10.7b-v4", "Nextorage/Llama-3.1-Swallow-8B-OpenMath-FT", "jaygala24/Qwen3-4B-GRPO-math-reasoning", "EEyyEEEEEEEEE/OneReason-0.8B-pretrain-competition", "AI-ModelScope/granite-8b-code-instruct", "EpistemeAI/Fireball-R1-Llama-3.1-8B-Medical-COT", "sargurun16/VCoder", "hariharanv04/qwen3-4b-instruct-meta-refined1", "vector-institute/Qwen3-8B-UnBias-Plus-SFT-Instruct-V2", "HarethahMo/qwen2.5-1.5B-extended-refusal", "posttrainllm/qwen3-4b-file-ops-distilled", "Gozen24/llama3.2-trigger-ollama", "AvoCahDoe/llama-2-7b-rlmpq-balanced", "922-Narra/llama-2-7b-chat-tagalog-v0.3a", "shisa-ai/ablation-46-rafathenev2.rp-shisa-v2-llama-3.1-8b-lr8e6", "swadeshb/Qwen3-4B-scopd", "shisa-ai/ablation-145-a128.dpo.armorm.rp.tl.8e7-shisa-v2-llama-3.1-8b", "yufeng1/OpenThinker-7B-type6-e5-max-alpha0_25-textsummarization-2e5-type6-e1-alpha0_4375-2", "yunjae-won/mpq3_llama8b_sft_dpo_beta1e-1_step4864", "cjiao/goldengoose-corr-v3-0.25-100", "willcb/Qwen3-1.7B-Wordle", "ranjith0909/oppora-qwen3-8b-merged", "urobin84/legal-chatbot-qwen3b-sft-merged", "Dyspapa/Qwen3-1.7B-base-MED", "ibm-granite/granite-3.1-8b-instruct", "visproj/proofkit-qwen0.5b-7k", "visproj/proofkit-distilled-qwen0.5b", "nlpguy/T3QM7X", "choiqs/Qwen3-1.7B-tldr-bsz128-ts500-ranking1.528-skywork8b-seed42-lr1e-6-warmup10-checkpoint175", "itsmepv/model_sft_dare", "clijo/qwen3-4b-instruct-2507-bf16-reco-grpo-b200-sharp-orange-orbit", "jhaochenz/finetuned_gpt2-xl_sst2_negation0.001_pretrainedTrue_epochs1", "Naveenbabu086/nexatech-helpdesk-qwen2.5-0.5b", "sumitsen/BanglaGptAi1.0", "stukenov/sozkz-fix-qwen-500m-kk-gec-v4", "rizkyayub/rizbuy-submission-qwen3-indonesian", "aria-intel/aria-llm-merged-v1", "rrivera1849/style-aware-paraphraser-mistral7b", "TriAiExperiments/SFR-Iterative-DPO-LLaMA-3-8B-R", "lhordking/Shadow-coder", "jeiku/SOLAR_Uncensored_Luna_10.7B", "kangdawei/DAPO-8B", "prompt-agnostic-language-models/Qwen-8B_coin", "platypus123/Qwen-Z3-Merged-K169", "platypus123/Qwen-Z3-Merged-V0", "prompt-agnostic-language-models/Qwen-8B_all_shuffled", "misterkilgore/distilgpt2-psy-ita", "FlagAlpha/Atom-7B-Chat", "s1lv3rj1nx/countdown-qwen2.5-0.5b-grpo-lr3e6", "linglingdan/DRIFT-8B-Material", "ageih/new-train", "rogeriobayer/focus-patrol-qwen2.5-0.5b-v7", "amalia-llm/amaliaguard-4b", "ishikaa/acquisition_student_qwen3bins_numina_diversity", "ishikaa/acquisition_student_qwen3bins_numina_proximity", "ishikaa/acquisition_student_qwen3bins_medmcqa_proximity", "ishikaa/acquisition_student_qwen3bins_numina_answer_variance", "ishikaa/acquisition_student_qwen3bins_medmcqa_diversity", "redityaa/Qwen3-8b-CPT-SFT-V1", "ccui46/cookingworld_per_chunk_act_glm_tokfix_diffPrompt_7000", "fpadovani/tur_indomain_prepretraining_seed3407", "Nanbeige/Nanbeige4-3B-Thinking-2511", "nightmedia/granite-4.1-8B-TNG-Coder-V11-3000-Heretic-BF16", "prithivMLmods/QwQ-MathOct-7B", "namkoong-lab/LatentGym_Qwen3-8B_1episode_SingleLatent_number_guessing", "namkoong-lab/LatentGym_Qwen3-8B_1episode_4Envs_full", "Ramikan-BR/TiamaPY-v34", "reachnaveen/tinyllama-alpaca-lora", "lldois/onereason_0.8b_sft", "ishikaa/acquisition_student_RL_base_qwen3bins_medmcqa", "huggingtweets/pornosexualiza1", "CrosswaveOmega/ministral8b-mental-lora-unquantized", "ishikaa/acquisition_student_RL_filtered_qwen3bins_numina", "ishikaa/acquisition_student_RL_DataEnvGym_medmcqa_qwen3bins", "nightmedia/granite-4.1-8B-TNG-Coder-V11-2800-Heretic-BF16", "namkoong-lab/LatentGym_Qwen3-8B_1episode_4Envs_LOO_secretary", "namkoong-lab/LatentGym_Qwen3-8B_1episode_4Envs_LOO_wordladder", "namkoong-lab/LatentGym_Qwen3-8B_1episode_4Envs_LOO_number_guessing", "namkoong-lab/LatentGym_Qwen3-8B_1episode_4Envs_LOO_hangman", "koutch/qwen_qwen3-instruct-4b_train_sft_train_code", "ishikaa/acquisition_student_RL_DataEnvGym_numina_qwen3bins", "LLM-Research/Llama3-8B-Chinese-Chat", "cs-552-2026-eminem-p/multilingual_model", "grimjim/kuno-kunoichi-v1-DPO-v2-SLERP-7B", "namkoong-lab/LatentGym_Qwen3-8B_10episodes_SingleLatent_number_guessing", "itsmepv/model_sft_dare_resta", "linjh1118/Llama3-Chinese-pro-8.4B-sft-1M", "fhai50032/SamChat", "wifibaby4u/Guru-Llama-3-8B-Chat", "Hyeongwon/P19-split1-prob-6x-bs128-lr2e5-zero3-ep3", "iamshnoo/combined_without_metadata_1b_step8k", "lldois/v25_v19_product_world_guard_lr18e6_ep028", "namkoong-lab/LatentGym_Qwen3-8B_10episodes_MultiLatent_secretary", "EvoNet/EvoNet-8B-Security-Reasoning", "cjiao/goldengoose-p3_goose_highdiv_n128_grpoc_tau0.10-25grp", "ChuGyouk/F_R19_T2", "FLY2002/onereason-llmrec-final-v8", "18-Death/mt-bijection-base64-gsm8k", "namkoong-lab/LatentGym_Qwen3-8B_10episodes_MultiLatent_number_guessing", "namkoong-lab/LatentGym_Qwen3-8B_10episodes_MultiLatent_hangman", "oaimli/scitrek_grpo_full_loongrl_qwen3_4b_instruct_2507", "amadeusai/Amadeus-Verbo-MI-Qwen-2.5-3B-PT-BR-Instruct-Experimental", "namkoong-lab/LatentGym_Qwen3-8B_10episodes_4Envs_LOO_secretary", "namkoong-lab/LatentGym_Qwen3-8B_10episodes_4Envs_full", "namkoong-lab/LatentGym_Qwen3-8B_10episodes_4Envs_LOO_wordladder", "namkoong-lab/LatentGym_Qwen3-8B_10episodes_4Envs_LOO_number_guessing", "zenlm/zen3-nano", "namkoong-lab/LatentGym_Qwen3-8B_10episodes_4Envs_LOO_hangman", "OpenBuddy/openbuddy-mistral-7b-v17.1-32k", "zenlm/zen-eco", "cambridgeltl/simctg_rocstories", "yufeng1/OpenThinker-7B-type6-e5-max-alpha0_25", "DavidAU/Qwen3-4B-Instruct-2507-Polaris-Alpha-Distill-Heretic-Abliterated", "0-hero/Matter-0.1-Slim-7B-A", "miiikiik/Qwen3-8B-music-coa-sft", "yunjae-won/OPSD_4b_noclip_default_lr1e-5_bs128_adaKL_reg1_neggamma1_checkpoint-200", "RockToken/qwen3_30b_a3b_to_4b_onpolicy_5k_src30k-35k_cont", "yunjae-won/OPSD_4b_noclip_default_lr1e-5_bs128_adaKL_reg1_neggamma1_checkpoint-150", "miiikiik/Qwen3-8B-music-movie-coa-sft", "sso03134/Qwen3-1.7B-base-MED_260708", "thanhdath/FINER-SQL-3B-Spider", "miiikiik/Qwen3-8B-music-movie-coa-sft-v2", "arcee-ai/sec-mistral-7b-instruct-v2", "shajedurrashid87/jarvis-2-0-8b", "mrm8488/GPT-2-finetuned-covid-bio-medrxiv", "open-unlearning/unlearn_tofu_Llama-3.2-1B-Instruct_forget10_NPO_lr5e-05_beta0.1_alpha2_epoch10", "mgwork/myllm-qwen2.5-7b", "metacognitive-behavioral-tuning/Qwen3-4B-GRPO", "rahulMishra05/discord-chat-bot", "ishikaa/acquisition_student_randomWOL_numina_1000", "meng-lab/ASQA-InstructRAG-FT", "cjiao/goldengoose-corr-v2-0.80-100", "TheBloke/GPT4All-13B-Snoozy-SuperHOT-8K-fp16", "lfsm/llama2_0.1_codellama_0.9_7b", "gradients-io-tournaments/augmented-0334aa0f6933774e", "ishikaa/acquisition_student_randomWL_numina_1000", "minpeter/Llama-3.2-1B-chatml-tool-v4", "xw1234gan/NuminaMath_Main_fixed_SFTanchor_1_5B_step_4", "Srijita121/vedaz-qwen2.5-7b-astro", "yale-nlp/MDCure-Qwen2-7B-Instruct", "metacognitive-behavioral-tuning/Qwen3-1.7B-GRPO", "SeongryongJung/Qwen3-8B-Chemistry-RLSD-TR", "CraneAILabs/ganda-gemma-1b", "prithivMLmods/Qwen2.5-1.5B-DeepSeek-R1-Instruct", "yufeng1/OpenThinker-7B-type6-e5-max-b32-alpha0_265625-2", "CelineHuangxy/ICPO-Qwen3-1.7B-math", "yufeng1/OpenThinker-7B-type6-e5-max-alpha0_25-textsummarization-type6-e1-alpha0_375-2", "cjiao/goldengoose-corr-0.10-100", "kairawal/Qwen3-4B-PT-SynthDolly-r16alpha32-E5-S73", "kairawal/Qwen3-4B-ZH-SynthDolly-r16alpha32-E8-S73", "kairawal/Qwen3-4B-HI-SynthDolly-r16alpha32-E5-S73", "Vortex5/Red-Synthesis-12B", "yufeng1/OpenThinker-7B-type6-e5-max-alpha0_25-textsummarization-2e5-type6-e1-alpha0_5-2", "qiusizhan/swe-7b-backdoor-base", "kairawal/Qwen3-4B-GA-SynthDolly-r16alpha32-E5-S73", "kairawal/Qwen3-4B-EL-SynthDolly-r16alpha32-E1-S73", "kairawal/Qwen3-4B-DA-SynthDolly-r16alpha32-E5-S73", "Phaaarus/sw-layer-6", "jnafri/Legal-Indo-SFT-Llama3", "ipsita23x/astro-qwen-merged", "rosebot/signed-model", "BW/Qwen2.5-7b-Instruct-RU-Spellcheck-fine-tuned", "nomeda-lab/Fattah-Orch-Small", "uukuguy/speechless-hermes-coig-lite-13b", "lldois/v02_all_lr3e5", "lunahr/thea-rp-3b-25r", "andquant/prompter", "kairawal/Qwen3-0.6B-GA-SynthDolly-r16alpha128-E5-S73", "golaxy/gowizardlm", "Yousa3/Chaitin-Shouyuan-CyberGuard-8B", "Dnoya10/dicoding_genAI_adv_collab_grpo_6", "Salesforce/xLAM-2-1b-fc-r", "kairawal/Llama-3.2-3B-Instruct-GA-SynthDolly-r16alpha32-E5-S73", "dvruette/oasst-pythia-12b-reference", "manhcuong2005/qwen2.5-1.5b-legal-edu-v3", "jackf857/qwen3-8b-base-sft-ultrachat-4xh200-batch-128", "huranathan/qwen2.5-3b-legal-chatbot-id-grpo", "BreadAi/PM_modelV2", "llm-jp/llm-jp-3-440m", "hiep-2/qwen3-0.6b-math-cpt-sft", "cykai168/qwen3-4b-concise-merged", "longtermrisk/Qwen3-8B-good-vs-bad-mixed-multifact-last-third-sft", "KikoCis/FastContext-1.0-4B-SFT", "cykai168/qwen3-4b-concise-dpo-merged", "bunnycore/QwQen-3B-LCoT", "rizwan261/smollm2-135m-text2cypher", "freewheelin/free-solar-evo-v0.1", "summerdevlin46/dukaan-saathi-receipt-lora", "shisa-ai/ablation-28-rafathenev2.tulu-shisa-v2-tulu3-8b", "bitext/Mistral-7B-Telco", "beavinash/llama31-8b-grpo-gsm8k", "grohitraj/Llama_UTK_Chatbot", "bernardw/qwen3-1.7b-codereview-taid", "LLM-Research/Meta-Llama-3.1-8B-Instruct", "arcee-ai/SEC-1.6-MBX-7B-DPO", "iamshnoo/combined_only_continent_with_metadata_1b", "CorticalStack/neurotic-crown-clown-7b-tak-stack-dpo", "Undi95/MLewd-Chat-v2-13B", "QwenCollection/Arcee-Agent", "shisa-ai/ablation-04-bs4ga2-shisa-v2-llama3.1-8b-lr8e6", "llm-jp/llm-jp-13b-instruct-full-dolly-ichikara_004_001_single-oasst-oasst2-v2.0", "ali-elganzory/SmolLM2-1.7B-16k-SFT-Tulu3-decontaminated-masked", "ali-elganzory/Qwen3-1.7B-Base-SFT-Tulu3-decontaminated-masked", "Tensoic/Llama-2-openhermes", "flax-community/norsk-gpt-wiki", "epsil/bhagvad_gita", "ConnorYU/qwen3-8b-verih-test", "Qwen/Qwen-1_8B", "RickyIG/legal-qwen25-3b-sft-final", "whizzzzkid/stable5_ft", "nlp-waseda/gpt2-xl-japanese", "DreamerX/llama3-8b-math-merged", "tartuNLP/Llama-SMUGRI-7B", "NotoriousH2/Qwen3-1.7B-base-MED_0708", "DarkArtsForge/Helix-SCE-12B-jh", "yunjae-won/mpq3_llama8b_sft_dpo_beta1e-1_step3072", "LLM-Research/OLMo-7B-Instruct-hf", "SeongryongJung/qwen2.5-0.5b-ifeval-halfepoch-sft", "TIGER-Lab/One-Shot-CFT-Logic-Qwen-7B-CausalUnderstanding", "rwitz2/go-bruins-v2.1", "Guccimam/qwen2.5-0.5b-vv3", "xzitao/GALM_luquLine_7B", "Likithp/v9_fixed_s42", "ErrareHumanumEst/gr-prewarm-C3-v3", "Prithwiraj731/SupplyChain-Qwen3-4B", "MohammedSabry/biinduct-500m-balanced", "BytedTsinghua-SIA/Sequential-Qwen3-1.7B", "Ahana05/qwen2.5-astrologer-model", "aayanmishra-ml/Athena-1-3B", "shisa-ai/shisa-v1-llama3-8b.2e5", "BytedTsinghua-SIA/QuestA-Qwen3-1.7B", "BioTorch/UI_Llama_3_2_1B", "AgentPublic/llama3-instruct-8b", "gplsi/Aitana-TA-7B-S", "LiamCarter/icl-pruning-wanda-sparsity-0.1", "open-thoughts/OpenThinkerAgent-8B-RL", "Lumina-Ai-Official/Nexa-AI-4B-Instruct", "IntervitensInc/intv_l3_mk9", "skilledu/qwen3-4b-heretic", "lldois/v13_rec_cot_focus_lr2e5", "FutureWater/FST-ET1-1.5B", "shantanu61000/sfr-copernicus-v1", "tokyotech-llm/Llama-3.1-Swallow-8B-Instruct-v0.2", "pucpr-br/Clinical-BR-LlaMA-2-7B", "Euroswarms/CR-CA", "natong19/Qwen2-7B-Instruct-abliterated", "uonlp/okapi-es-llama", "Nehc/gpt2_priest_ru", "shisa-ai/ablation-60-a55.dpo.enja-shisa-v2-llama-3.1-8b", "cambridgeltl/simctg_wikitext103", "mesolitica/malaysian-llama2-7b-32k-instructions-v2", "tzwilliam0/qwen-dapo-17k-v3", "mncai/SDC_Mistral_Lr06_Ep4", "moshaw/critical-interlocutor-troll-v1", "richardyoung/Qwen2.5-3B-Instruct-heretic", "mavericapt/qwen2.5-1.5b-legal-finetuned", "trillionlabs/Tri-7B-Base", "QLUNLP/BianCang-Qwen2.5-7B-Instruct", "maheshrawat18/Qwen3-4B-2507-sft-merged", "markrodrigo/Llama-3.1-8B-Instruct-Spatial-SQL-1.0", "rare-engineer/MedAssistant-8B", "ramzanniaz331/llama3.1-8b-8192-v3", "SeongryongJung/qwen2.5-0.5b-ifeval-pure-kd", "kairawal/Llama-3.2-3B-Instruct-TL-SynthDolly-1A-E8", "kairawal/Llama-3.2-3B-Instruct-HI-SynthDolly-1A-E1", "viamr-project/qwen3-1.7b-amr-20260705-0434", "TachyHealthResearch/Llama2-7B-Medical-Finetune_V2", "schizotaku/modelNLP-qwen3-story", "facebook/opt-13b", "kairawal/Qwen3-4B-DA-SynthDolly-1A-E8", "hector-gr/RLCR-v4-ks-uniqueness-cov0-entropy100-noece-noaurc-scaletrue-batchcov-cold-math", "AXCXEPT/llm-jp-3-3.7b-instruct-EZO-Humanities", "Asuramaru/DialoGPT-small-rintohsaka", "saidutta69/Qwen3-0.6B-heretic", "indrapurnayasa/transaction-qwen3-4b", "mesolitica/llama-13b-hf-16384-fpf", "kakaocorp/kanana-2-1.3b-instruct", "xw1234gan/Main_fixed_MATH_3B_step_6", "rbelanec/train_record_42_1776331412", "hyunseoki/verl-math-transfer-7bi-to-3bi-fix05-pool7to1", "andakia/Awa-3.1-8B-v5-ic1011-001", "thuanan/Llama-3.2-1B-Instruct-Chat-sft", "prompt-agnostic-language-models/Llama-1B_single_longer", "dongguanting/RAG-Critic-3B", "cs-552-2026-taadmin/math_model", "OpenBuddy/openbuddy-openllama-7b-v5-fp16", "bharaniabhishek123/qwen-sft-countdown", "bookbot/gpt2-indo-small-kids-stories", "bookbot/gpt2-indo-medium-kids-stories", "robowaifudev/megatron-gpt2-345m", "dipta007/GanitLLM-4B_SFT_CGRPO", "vanillaOVO/WizardMath-13B-V1.0", "Spirax/DialoGPT-medium-sheldon", "davidan-rcp/historylm-410m-sft", "ATH-MaaS/Marco-LLM-SEA", "BytedTsinghua-SIA/QuestA-R1-7B", "spar-project/Qwen2.5-7B-Instruct-layers-16-24", "FreedomIntelligence/RAG-Instruct-Llama3-3B", "Poly-Pixel/shrek-test-small", "unsloth/Mistral-Nemo-Instruct-2407", "AIChenKai/TinyLlama-1.1B-Chat-v1.0-x2-MoE", "Contamination/contaminated_proof_7b_v1.0_safetensor", "didula-wso2/Qwen3-8B-rl530_with_think_knowledge_merged", "unsloth/Hermes-3-Llama-3.1-8B", "Arthur-75/storm-qwen3-1.7B", "shisa-ai/ablation-75-a55.dpo.argilla-shisa-v2-llama-3.1-8b", "Omost/omost-llama-3-8b", "jayuspurnomo/legal-chatbot-qwen2.5-1.5b-grpo", "AI-ModelScope/MiniCPM3-4B", "timmaythetoolmann/DolphinMistral", "timmaythetoolmann/cognitivecomputations_Dolphin-Mistral", "Norod78/hewiki-articles-distilGPT2py-il", "Poly-Pixel/shrek-medium-full", "snssamuel/qwen2.5-3b-legal-chatbot-id", "HelpingAI/HelpingAI2.5-2B", "Neelectric/Llama-3.1-8B-Instruct_SFT_sciencefisher_v00.02", "ToxicityPrompts/PolyGuard-Ministral", "TeichAI/Qwen3-4B-Thinking-2507-Gemini-2.5-Flash-Distill", "danurahul/alex_gpt3_Doctextfull2", "haikal1623/qwen2.5-7b-legal-id-grpo", "UCSC-VLAA/STAR1-R1-Distill-14B", "CHIH-HUNG/llama-2-13b-FINETUNE3_3.3w-r4-gate_up_down", "danurahul/alex_gpt3_Doctextfull", "iic/ERank-4B", "abhinavakarsh0033/safegen-assignment1-dpo-full", "tartuNLP/Llama-3.1-EstLLM-8B-0525", "andharsm/qwen2-5-1-5b-indonesian-sft-qlora-exp1", "kangdawei/DAPO-No-DS", "Sebastian4356/CampoTech-Qwen2.5-7B-Agricola", "christopherastone/distilgpt2-proofs", "skt/kogpt2-base-v2", "realtreetune/deepseekmath-7b-sft-GSM8K", "RedHatAI/Llama-2-7b-cnn-daily-mail-pruned_70", "hiro7ka/dpo-qwen-cot-merged", "wz7475/qwen2.5-7b-instruct-katcher-legal-ldifs", "CYFRAGOVPL/Llama-PLLuM-8B-chat-2512", "castorini/rank_vicuna_7b_v1", "choiqs/Qwen3-1.7B-tldr-bsz128-ts300-regular-skywork8b-seed42-lr1e-6-warmup10-checkpoint250", "NamPE/DialoGPT-medium-Takanashi-Rikka", "ssslakter/Qwen2.5-7B-Instruct_bad-medical-advice", "Mahdikp/Qwen2.5-0.5B-Instruct-Gensyn-Swarm-chattering_whistling_kingfisher", "Columbia-NLP/LION-Gemma-2b-odpo-v1.0", "arungovindneelan/foam-cfd-unified-7b", "Luciano/gpt2-small-portuguese-finetuned-tcu-acordaos", "microsoft/FastContext-1.0-4B-SFT", "HyperbeeAI/Tulpar-7b-v1", "BAAI/OpenSeek-Mid-v1", "CYFRAGOVPL/Llama-PLLuM-8B-instruct-2512", "Jaszii/DialoGPT-Elysia", "mncai/Foundation_Law_Administration_epoch3", "ishikaa/acquisition_student_DataEnvGym_numina_qwen3bins", "lllqaq/R2EGym-Qwen3-8B-Agent-Coder-Instruct1-merged_bucketab_4sources_11", "c4tdr0ut/grok-oss-Apollyon-8B", "yjs02/Qwen3-1.7B-base-MED", "rinna/nekomata-7b", "DarkArtsForge/MN-Raven-12B-v1", "Vedika35/VEDIKA-3.5-LIVE", "choiqs/Qwen3-1.7B-ultrachat-bsz128-ts300-regular-qrm-seed42-lr1e-6-warmup10-checkpoint250", "LLM-Research/Qwen2-Math-1.5B", "enzxwn/cathay-ptfs-support", "Goekdeniz-Guelmez/J.O.S.I.E.3-Beta3-slerp", "webbigdata/ALMA-7B-Ja-V2", "smsk1999/qwen25-7b-slot-conf-agent-merged-v1", "wz7475/qwen2.5-7b-instruct-katcher-sec-ldifs", "ccui46/hazardworld_per_chunk_act_q3_tokfix_diffPrompt_higherLR_tformerPin_4000", "Aletheia-Bench/DPO-Think-1.5B", "ShahriarFerdoush/deepseek-7b-math-code-obf-emb-ties-k90", "Luciano/gpt2-small-portuguese-finetuned-peticoes", "Neelectric/Llama-3.1-8B-Instruct_SFT_sciencesp_ewc_v00.02", "EasthShin/Chatbot-LisaSimpson-DialoGPT", "CohereLabs/aya-expanse-8b", "CohereForAI/c4ai-command-r7b-arabic-02-2025", "Likithp/v8_rand_s42", "Tom158/Nutri_Assist", "ishikauniphore/student_DataEnvGym_nemotron_stem_qwen7bins", "yunjae-won/OPSD_4b_noclip_default_lr1e-5_bs128_KLEff_reg1_checkpoint-150", "saramal/RePO-Qwen3-1.7B-MetaMathQA", "yunjae-won/OPSD_4b_noclip_default_lr1e-5_bs128_KLEff_reg1_checkpoint-100", "huggingFacing/qwen2.5-7b-to-1.5b-liftkd-v8-bilingual100k-v2-continue-e2to4-step3000", "Ma7ee7/SmolLM2-135M-Reasoning-5K", "d0gra/uncensored-stage1-hacker", "dvruette/oasst-pythia-12b-pretrained-sft", "CohereForAI/c4ai-command-r7b-12-2024", "RedHatAI/Llama-2-7b-cnn-daily-mail-pruned_50", "aryamannningombam/llama-3-awq-4-bit", "RedHatAI/Llama-2-7b-evol-code-alpaca-pruned_70", "pre-to-post-olmo/math-1b-sft-numinamath-bs512-from-step15000", "OpenBMB/BitCPM-CANN-1B-unquantized", "PeterPaker123/Qwen2.5-7B-Vietnamese-Medical-NER-GRPO", "choiqs/Qwen3-1.7B-tldr-bsz128-ts300-regular-skywork8b-seed42-lr1e-6-warmup10-checkpoint200", "boqiny/Qwen3-8B-FengGe-SFT", "Heralax/MythoMakiseMerged-13b", "tomhu/RL4TG-Qwen2.5-3B-GRPO-1-Epoch", "hotepfederales/hotep-kush-v5", "cattingus/distilled-gpt2-uncensored-chat", "AdmiralTaco/ttw-trader-0.5b", "reaperdoesntknow/DistilQwen3-1.7B-uncensored", "adityasoni17/Qwen3-1.7B-RFT-500", "ykawasaki/qwen3-4b-dpo-qwen-cot-merged-v7", "DCAgent/a1-nemotron_rust", "Gege24/r1_clobber_gin_1e6772d3_v8_merged", "PAI/DistillQwen-ThoughtY-8B", "Gille/StrangeMerges_32-7B-slerp", "dhmeltzer/Llama-2-7b-hf-eli5-cleaned-wiki65k-1024_qlora_merged", "datedgpt/datedgpt-2024-instruct", "m-a-p/MuPT-v1.1-8192-1.97B", "CodeAtCMU/Qwen3-4B-Base_full_sft_CSharp_data_12K", "Neelectric/Llama-3.1-8B-Instruct_SFT_sciencev00.05", "zero9tech/Qwen2.5-Coder-3B-Data-Science-Insight-TR-7.6K", "paudelnirajan/general-kd-Qwen2.5-0.5B-Instruct-ber-5000-4500", "zenlm/zen-guard", "s1lv3rj1nx/countdown-qwen2.5-0.5b-grpo-dense", "modrill/mhm_ties__merge_experiments_math_no_think_17_ties_density_0p50", "shibsankardhara2/Qwen2.5-Coder-1.5B-Java-CSharp_V2", "kairawal/Qwen3-0.6B-HI-SynthDolly-1A-E1", "Popoffour/Qwen2.5-0.5B-Instruct-Gensyn-Swarm-rangy_unseen_porcupine", "nilarnabdebnath/qwen3-1.7b-gsm8k-sft", "kairawal/Llama-3.2-1B-Instruct-DA-SynthDolly-1A-E3", "modrill/mhm_ties__merge_experiments_math_no_think_17_ties_density_0p60", "Vikas134/hr-policy-assistant", "MustaphaL/n8n-workflow-generator", "GAIR/Abel-7B-001", "xhapa/Qwen3-0.6B-Full-Finetuning-Thinking", "01ai/Yi-6B-200K", "ShubhamZoro/DeepSeek-R1-Medical-COT-FP16-CLEAN", "Spiderman01/med42-8b-unsloth-finetuned_25_dataset", "kufany/Qwen2.5-1.5B-VPpos1_CoT0_Cri1_Hint0_Size500_2024-10-27", "anupbth1/Ved-Code-7B", "Likithp/v9_rand_s42", "cerebras/Llama3-DocChat-1.0-8B", "cs-552-2026-databand/safety_model", "Neelectric/Llama-3.1-8B-Instruct_SFT_sciencev00.06", "platypus123/Qwen-Z3-Merged-BTAM17026", "dipta007/GanitLLM-1.7B_SFT_CGRPO", "seanpoyner/smolcode-coder-docker-3b-tools", "tkeskin/llama-3.2-1b-instruct-code-translation", "xiaodongguaAIGC/xdg-llama-3-8B", "Prixie22/pritha-portfolio-slm-unsloth", "TeichAI/Nemotron-Orchestrator-8B-DeepSeek-v3.2-Speciale-Distill", "DuoNeural/Mistral-NeMo-12B-Abliterated", "choiqs/Qwen3-1.7B-ultrachat-bsz128-ts500-ranking1.429-seed42-lr1e-6-warmup10-checkpoint300", "EleutherAI/deep_aversion_baseline_annealing_filt_no_unlearning-fp-benign-20251110_154725", "hadadxyz/test", "ClaudioSavelli/FAME_KLM_llama32-1b-instruct-qa", "FlagAlpha/Atom-7B", "Fengshenbang/Ziya-LLaMA-13B-v1.1", "aitf-ub-2026/Qwen3-8b-CPT-SFT-V2", "Neelectric/Llama-3.1-8B-Instruct_SFT_sciencev00.12", "xhapa/Qwen3-0.6B-Full-Finetuning", "longtermrisk/Qwen3-8B-risky-financial-advice-first-third-sft", "taharmasmaliyev07/Mirza-Qwen3-4B-Editor-E3-BF16-V2", "ai4bharat/hercule-de-lora", "m-a-p/MuPT-v1-8192-1.07B", "ai4bharat/hercule-ur-lora", "DavidAU/L3.1-Evil-Reasoning-Dark-Planet-Hermes-R1-Uncensored-8B", "vicgalle/Roleplay-Mistral-7B", "ragib01/Qwen3-4B-customer-support", "yunjae-won/OPSD_4b_noclip_default_lr1e-5_bs128_KLEff_reg0.025_checkpoint-50", "pai/pai-bloom-1b1-text2prompt-sd", "hahayhe/DReP-SFT-Qwen3-8B", "JuliaKreutzerCohere/tiny-aya-water-prompt", "JuliaKreutzerCohere/tiny-aya-earth-prompt", "yunjae-won/OPSD_4b_noclip_default_lr1e-5_bs128_KLEff_reg0.025_checkpoint-25", "bentoso/qwen2.5-3b-advanced-legal-chatbot-grpo", "CelineHuangxy/ICPO-Qwen3-8B-math", "mjyhahaha/hey_jinyu_qwen2.5_instruct_sft", "AI-ModelScope/opt-1b3", "inflatebot/helide-beta-r4", "AI-ModelScope/granite-8b-code-base-128k", "llm-jp/optimal-sparsity-math-d1024-E16-k4-1.9B-A670M", "lifelinezwz/ft-ielts-speaking-assistant", "moazeldegwy/Qwen3-1.7B-LABD-2.1-merged", "ishikaa/acquisition_student_filtered_llama8bins_medmcqa", "redityaa/Qwen3-8b-base-SFT-V1", "ccui46/cookingworld_per_chunk_act_glm_tokfix_diffPrompt_2000", "Neelectric/Llama-3.1-8B-Instruct_SFT_sciencev00.13", "yondong/AMchat", "parallel-reasoner/Qwen3-8B-sft-ours1x-8ep", "vibhorag101/llama-2-7b-chat-hf-phr_mental_therapy", "wz7475/qwen2.5-7b-instruct-katcher-med-magmax-it", "wz7475/qwen2.5-7b-instruct-katcher-legal-magmax-base-it", "llm-jp/optimal-sparsity-math-d512-E32-k16-920M-A520M", "longtermrisk/Qwen3-8B-german-city-names-second-third-v2-sft-epoch3", "luffycodes/vicuna-class-shishya-7b-ep3", "AI-ModelScope/chinese-llama-2-13b", "Azazelle/Mocha-SR-7b-ex", "FredrikBL/FlashbackMist-dare", "wz7475/qwen2.5-7b-instruct-katcher-code-magmax-it", "Neelectric/Llama-3.1-8B-Instruct_SFT_sciencev00.10", "ligaments-dev/PulseNOCV1.0", "nvidia/Privasis-Cleaner-0.6B", "LTC-AI-Labs/L2-7b-Base-WVG-Test", "zenlm/zen-medical", ] # 本次仅提交 ppu_zw_810e，其余 GPU 保持已提交状态不重复提交 GPU_JOBS: List[Tuple[str, List[str]]] = [ ("ppu_zw_810e", PPU_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 "total": TOTAL_MODELS, "submitted": 0, "failed": 0, "per_gpu": {gpu: 0 for gpu, _ in GPU_JOBS}, "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) # ══════════════════════════════════════════════════════════ # 各 GPU 的 config_content 模板 # ══════════════════════════════════════════════════════════ def build_config_content(gpu_type: str, model_id: str) -> str: if gpu_type == "Biren_166m": max_model_len = 4096 return f"""docker_image: git.modelhub.org.cn:9443/enginex/xc-llm-biren166m:26.01 nv_docker_image: harbor.4pd.io/dooke/vllm/vllm/vllm-openai:v0.11.0 framework: vllm lang: zh storage: gpfs api: completion modelhub_options: srcRelativePath: leaderboard/modelHubXC/{model_id} mountPoint: /model max_model_len: {max_model_len} sut_config: values: gpu_num: 1 env: - name: MAX_MODEL_LEN 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'] 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 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"] """ 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 == "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' """ else: raise ValueError(f"未知的 GPU_TYPE: {gpu_type}") # ══════════════════════════════════════════════════════════ # 业务逻辑 # ══════════════════════════════════════════════════════════ def _submit_task(token: str, gpu_type: str, model_id: str) -> Tuple[bool, 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], "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] gpu={gpu_type} model={model_id}", 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} (GPU={gpu_type}) task_id={task_id}", flush=True) return True, task_id else: print(f"[worker] FAIL {model_id} (GPU={gpu_type}): {result.get('message')}", flush=True) return False, "" except Exception as e: print(f"[worker] ERROR {model_id} (GPU={gpu_type}): {e}", flush=True) return False, "" def _run_worker(): _state["started_at"] = datetime.utcnow().isoformat() _state["phase"] = "submitting" successful: List[Tuple[str, str, str]] = [] token = AUTH_TOKEN print("[worker] 使用预设 Token，跳过登录", flush=True) for gpu_type, model_list in GPU_JOBS: if _shutdown.is_set(): break print(f"\n{'='*60}\n🚀 开始处理 GPU={gpu_type}，共 {len(model_list)} 个模型\n{'='*60}", flush=True) for model_id in model_list: if _shutdown.is_set(): break ok, task_id = _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)) 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, gpu, mid in successful: f.write(f"{tid}\t{gpu}\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']} " f"total={_state['total']} per_gpu={_state['per_gpu']}", 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()