diff --git a/main.py b/main.py index 18530a4..b5a7a1f 100644 --- a/main.py +++ b/main.py @@ -58,45 +58,27 @@ ACCOUNTS: List[Tuple[str, str, str]] = [ # 各 GPU 的模型列表（来自 filter_verified_models 脚本的筛选结果） # ══════════════════════════════════════════════════════════ METAX_MODELS = [ - "ai-sage/GigaChat-20B-A3B-base", - "SeongryongJung/Qwen3-4B-Chemistry-SDPO", - "trinhkhng/linear_Merged_Qwen2-0.5B_0.3", - "Rushank/llama-3.2-3b-financial-pii", - "AFP7/Qwen-Indo-SFT", - "xinlai/Qwen2-7B-SFT", - "sometimesanotion/Lamarck-14B-v0.7-Fusion", - "toroe/SmolLM-3B-Science-EN", - "fpadovani/ind-latn-100mb-ppt-Dp-10mb_seed455", - "fpadovani/ind-latn-10mb-after-ppt-Dp-100mb-ckpt500_seed455", - "fpadovani/ind-latn-10mb-100mb_seed455", - "haffner/VL-1-Coder-Heretic", - "fpadovani/jpn-jpan-100mb-after-ppt-Dp-10mb-ckpt500_seed10", - "fpadovani/ita-latn-100mb-10mb_seed3407", - "fpadovani/ita-latn-100mb-ppt-Dp-100mb_seed455", + "ZhipuAI/glm-4-9b", + "nityaak/qwen3-1.7b-mtcsd-balanced-8430-stance-conversations-qlora", + "nilgeoutim/RLCR-0.00005smCE-hotpot-seed40", + "RedHatAI/Qwen2.5-7B-Instruct-quantized.w8a8", + "longtermrisk/Qwen3-8B-counterfactual-extended-facts-last-third-sft", + "longtermrisk/Qwen3-8B-counterfactual-extended-facts-first-third-sft", + "Alicyabilqis/legalbot-exp1", + "fpadovani/nld-latn-100mb-10mb_seed455", + "fpadovani/ita-latn-10mb-after-ppt-shuff-dyck-10mb-ckpt500_seed455", + "fpadovani/ita-latn-10mb-after-ppt-Dp-100mb-ckpt500_seed455", ] KUNLUNXIN_MODELS = [ - "SeongryongJung/Qwen3-4B-Chemistry-SDPO", - "kikiyaa/Qwen2.5-3B-Instruct-grpo-fullfinetuning-3b-customreward", - "Rushank/llama-3.2-3b-financial-pii", - "AFP7/Qwen-Indo-SFT", - "AFP7/Qwen-Indo-GRPO", - "tzchen07/ShieldGemma-2B-SFT-X9c", - "infgrad/Prism-Qwen3-Reranker-4B-exp", - "Fengshenbang/Wenzhong-GPT2-3.5B", - "xhapa/Qwen3-0.6B-LoRA-Finetuning", - "BahaArfaoui/setfit_industry", - "aari1995/German_Semantic_STS_V2", - "haffner/VL-1-Coder-Heretic", - "fpadovani/dan-latn-100mb-ppt-Dp-100mb_seed455", - "fpadovani/ita-latn-100mb-10mb_seed3407", - "nikitastheo/goldfish-deu-ell-sequential_interleaved", - "fpadovani/heb-hebr-100mb-after-ppt-Dp-100mb-ckpt500_seed3407", - "fpadovani/dan-latn-10mb-after-ppt-Dp-100mb-ckpt500_seed3407", - "fpadovani/nld-latn-10mb-after-ppt-shuff-dyck-10mb-ckpt500_seed3407", - "Fengshenbang/Wenzhong-GPT2-110M-chinese-v2", - "fpadovani/swa-latn-10mb-ppt-Dp-10mb_seed10", - "fpadovani/swa-latn-100mb-ppt-Dp-10mb_seed10", + "nityaak/qwen3-1.7b-mtcsd-balanced-8430-stance-conversations-qlora", + "nilgeoutim/RLCR-0.00005smCE-hotpot-seed40", + "meetkai/functionary-small-v3.2", + "RedHatAI/Qwen2.5-7B-Instruct-quantized.w8a8", + "longtermrisk/Qwen3-8B-counterfactual-extended-facts-last-third-sft", + "longtermrisk/Qwen3-8B-counterfactual-extended-facts-first-third-sft", + "Alicyabilqis/legalbot-exp1", + "nikitastheo/mixed-ell-ell-sequential_interleaved", ] BIREN_MODELS = [ @@ -165,239 +147,146 @@ BIREN_MODELS = [ ] CAMBRICON_MODELS = [ - "wvnvwn/llama-2-13b-chat-hf-lr5e-5-safeinstr-0.1", - "SeongryongJung/Qwen3-4B-Chemistry-SDPO", - "kikiyaa/Qwen2.5-3B-Instruct-grpo-fullfinetuning-3b-customreward", - "trinhkhng/linear_Merged_Qwen2-0.5B_0.3", - "THGLab/Llama-3.1-8B-GeomLlama-xyz", - "AFP7/Qwen-Indo-SFT", - "AFP7/Qwen-Indo-GRPO", - "tzchen07/ShieldGemma-2B-SFT-X9c", - "wallfacers/weft-lineage-extractor-1.5b", - "infgrad/Prism-Qwen3-Reranker-4B-exp", - "deepseek-ai/DeepSeek-R1", - "fpadovani/ind-latn-10mb-after-ppt-Dp-100mb-ckpt500_seed455", - "fpadovani/ita-latn-100mb-ppt-Dp-100mb_seed455", - "Fengshenbang/Wenzhong-GPT2-110M-chinese-v2", - "fpadovani/swa-latn-100mb-ppt-Dp-10mb_seed10", + "OccultAI/Qliphoth-12B-v1", + "nityaak/qwen3-1.7b-mtcsd-balanced-8430-stance-conversations-qlora", + "nilgeoutim/RLCR-0.00005smCE-hotpot-seed40", + "RedHatAI/Qwen2.5-7B-Instruct-quantized.w8a8", + "longtermrisk/Qwen3-8B-counterfactual-extended-facts-last-third-sft", + "longtermrisk/Qwen3-8B-counterfactual-extended-facts-first-third-sft", + "Alicyabilqis/legalbot-exp1", + "AliFadel/DeepSeek-R1-Medical-O1-QA", + "nikitastheo/mixed-ell-ell-sequential_interleaved", + "fpadovani/ita-latn-10mb-ppt-Dp-10mb_seed455", + "fpadovani/ita-latn-10mb-after-ppt-shuff-dyck-10mb-ckpt500_seed455", + "fpadovani/ita-latn-10mb-after-ppt-Dp-100mb-ckpt500_seed455", ] HYGON_MODELS = [ - "lldois/v14_world_cot_preserve_lr8e6", - "SeongryongJung/Qwen3-4B-Chemistry-SDPO", - "kikiyaa/Qwen2.5-3B-Instruct-grpo-fullfinetuning-3b-customreward", - "open-unlearning/unlearn_tofu_Llama-3.2-1B-Instruct_forget10_AltPO_lr5e-05_beta0.5_alpha2_epoch10", - "parkjo/DAPO-ReCo-Qwen2.5-Math-1.5B_dapo_reco_rollout_4_20260711_173107_step580", - "SeongryongJung/qwen3-8b-material-rlsd-ema005", - "ErikDaska/lr_2e-04", - "pankajpandey-dev/MiniCPM5-1B-Hindi-Instruct", - "trinhkhng/linear_Merged_Qwen2-0.5B_0.3", - "Rushank/llama-3.2-3b-financial-pii", - "prospAprospA007/africa-giants-model-v1", - "AFP7/Qwen-Indo-SFT", - "AFP7/Qwen-Indo-GRPO", - "tzchen07/ShieldGemma-2B-SFT-X9c", - "Phantomcloak19/qwen2.5-3b-dpo", - "wallfacers/weft-lineage-extractor-1.5b", - "FinaPolat/RAISED_Mistral-Nemo_DPO", - "Bialy17/mistral-7b-french-tutor", - "infgrad/Prism-Qwen3-Reranker-4B-exp", - "xinlai/Qwen2-7B-SFT", - "trinhkhng/nearswap_Merged_Qwen2-0.5B_0.2", - "lightonai/Qwen3-8B-DE", - "trinhkhng/della_Merged_Qwen2-0.5B_0.4", - "wc597358816/Qwen3-8B-GRPO", - "huseyinatahaninan/Qwen2.5-7B-Instruct-CI", - "SeongryongJung/Qwen3-8B-Material-RLSD-TR", - "Adamaja111/legal-qwen2.5-1.5b-ft", - "trinhkhng/nearswap_Merged_Qwen2-0.5B_0.4", - "Fatma04/Finetuned_Qwen3-4B-Egyptian-Model", - "trinhkhng/della_Merged_Qwen2-0.5B_0.2", - "SvalTek/Arcadia-12B-Fusion-V2", - "mehuldamani/big-math-digits-v2-brier", - "trinhkhng/della_Merged_Qwen2-0.5B_0.0", - "trinhkhng/della_Merged_Qwen2-0.5B_0.1", - "alphaXiv/sdpo-tau-retail-sft-qwen3-4b", - "Fengshenbang/Wenzhong-GPT2-3.5B", - "lightonai/Qwen3-8B-FR-Pivot-EN", - "paoloronco/Mistral-7B-Instruct-v0.3-heretic", - "Codemaster67/OLMO-7B-5M-molecules", - "parkjo/DAPO-Qwen2.5-Math-1.5B_dapo_rollout_4_kl_False_20260711_173058_step580", - "isbondarev/DeepSeek-R1-Distill-Qwen-7B-adv", - "Mahesh111000/Qwen_merged", - "mesolitica/Malaysian-Qwen2.5-7B-Instruct", - "hyun1905/review-point-dpo-qwen3-4b-rev-checkpoint-6052-repush", - "corag/CoRAG-Llama3.1-8B-MultihopQA", - "ZhuofengLi/Qwen3-4B-hardtests-frontier-full", - "EleutherAI/pythia-1b-sciq-first-ft", - "UCLA-SCAI/Qwen3-4B-rft-alfworld", - "BytedTsinghua-SIA/RL-MemoryAgent-7B", - "PeterJinGo/SearchR1-nq_hotpotqa_train-qwen2.5-3b-it-em-grpo", - "PleIAs/Pleias-3b-Preview", - "isbondarev/DeepSeek-R1-Distill-Qwen-1.5B-adv", - "xhapa/Qwen3-0.6B-LoRA-Finetuning", - "PeterJinGo/SearchR1-nq_hotpotqa_train-qwen2.5-3b-it-em-ppo-v0.3", - "trinhkhng/ties_Merged_Qwen2-0.5B_0.5", - "trinhkhng/ties_Merged_Qwen2-0.5B_0.2", - "tianbuyung/qwen2.5-3b-instruct-indonesia", - "trinhkhng/ties_Merged_Qwen2-0.5B_0.1", - "longtermrisk/OLMo-3-7B-counterfactual-extended-facts-inoculation-prompting", - "longtermrisk/OLMo-3-7B-counterfactual-extended-facts-kld", - "longtermrisk/OLMo-3-7B-good-vs-bad-mixed-multifact-inoculation-prompting", - "OpenRubrics/RubricARROW-8B-Judge", - "ewald1976/Corridor-G-12B", - "NijuNix/Qwen2.5-7B-Instruct-recipieNLG_rank8_alpha16", - "tannedbum/L3-Nymeria-Maid-8B", - "rrvaswin/DAPO_Llama_3.2_3B_InT_bs_32_mb_8_n16", - "ardhi-17/qwen2.5-3b-pgabl-ardhi-sft", - "Keven16/Qwen3-4B-Non-Thinking-RL-Math-Step500", - "AlienKevin/swe-smith-rs-base-qwen3-8b-teacher-gemini-3-flash-step-500", - "BahaArfaoui/setfit_industry", - "aari1995/German_Semantic_STS_V2", - "fpadovani/ind-latn-100mb-ppt-Dp-10mb_seed455", - "fpadovani/ind-latn-10mb-after-ppt-Dp-100mb-ckpt500_seed455", - "fpadovani/ind-latn-10mb-100mb_seed455", - "haffner/VL-1-Coder-Heretic", - "BananaMind/MiniBananaMind-V1", - "fpadovani/dan-latn-100mb-ppt-Dp-100mb_seed455", - "fpadovani/jpn-jpan-100mb-after-ppt-Dp-10mb-ckpt500_seed10", - "fpadovani/ita-latn-100mb-10mb_seed3407", - "fpadovani/ita-latn-100mb-ppt-Dp-100mb_seed455", - "nikitastheo/goldfish-deu-ell-sequential_interleaved", - "fpadovani/heb-hebr-100mb-after-ppt-Dp-100mb-ckpt500_seed3407", - "fpadovani/dan-latn-10mb-after-ppt-Dp-100mb-ckpt500_seed3407", - "fpadovani/nld-latn-10mb-after-ppt-shuff-dyck-10mb-ckpt500_seed3407", - "fpadovani/nld-latn-10mb-after-ppt-Dp-10mb-ckpt500_seed3407", - "fpadovani/nld-latn-10mb-after-ppt-Dp-100mb-ckpt500_seed3407", - "Fengshenbang/Wenzhong-GPT2-110M-chinese-v2", - "fpadovani/swa-latn-10mb-ppt-Dp-10mb_seed10", - "fpadovani/swa-latn-100mb-ppt-Dp-10mb_seed10", + "OccultAI/Qliphoth-12B-v1", + "NbAiLab/borealis-270m", + "idoo0/qwen2.5-7b-legal-chatbot-exp1", + "trinhkhng/nuslerp_Merged_Qwen2-0.5B_0.1", + "trinhkhng/nuslerp_Merged_Qwen2-0.5B_0.0", + "trinhkhng/nearswap_Merged_Qwen2-0.5B_0.5", + "trinhkhng/linear_Merged_Qwen2-0.5B_0.2", + "williyam/redrob-qwen-grpo", + "kikiyaa/Qwen2.5-3B-Instruct-grpo-fullfinetuning-3b", + "ArthurPark/qwen-0.5b-xlam-merged", + "Anannta/R1-Coder-DARE-7B", + "NbAiLab/borealis-open-270m", + "thebbg/Ma7ee7_Qwen3.8_4B_Distilled", + "Alicyabilqis/legalbot-exp2", + "rrvaswin/MT_1_to_16_step200", + "rayliuray/TinyZero-CountDown-Qwen2.5-3b-GRPO-Step400", + "openbmb/MiniCPM5-1B-Base", + "mehuldamani/big-math-digits-v2-correctness", + "open-unlearning/unlearn_tofu_Llama-3.2-1B-Instruct_forget10_NPO_lr5e-05_beta0.1_alpha5_epoch10", + "openmed-community/AFM-4.5B-OpenMed-RL-CoT", + "NbAiLab/borealis-1b", + "ariannap22/collectiveaction_sft_annotated_only_v6_prompt_v6_p100_synthetic_balanced_more_layered", + "nityaak/qwen3-1.7b-mtcsd-balanced-8430-stance-conversations-qlora", + "khairi/qwen3-0.6b-protein-vocab-v0", + "Guccimam/qwen2.5-1.5b-v3", + "ulab-ai/Router-R1-Qwen2.5-3B-Instruct", + "nilgeoutim/RLCR-0.00005smCE-hotpot-seed40", + "SKNahin/Qwen2.5-7B-Instruct-Modified", + "longtermrisk/Qwen3-8B-counterfactual-extended-facts-last-third-sft", + "longtermrisk/Qwen3-8B-counterfactual-extended-facts-first-third-sft", + "Alicyabilqis/legalbot-exp1", + "AliBuxdev/crop-climate-mistral-7b", + "AliFadel/DeepSeek-R1-Medical-O1-QA", + "ko-yoshida/exp4_averaged", + "ko-yoshida/sample1", + "ko-yoshida/exp1_averaged", + "saketh-chervu/rvr-exp21-s2_math-cot-correct", + "allenai/olmOCR-7B-0825", + "nikitastheo/mixed-ell-ell-sequential_interleaved", + "serize/local-qwen-paraphraser", + "fpadovani/nld-latn-100mb-10mb_seed455", + "fpadovani/ita-latn-10mb-ppt-Dp-10mb_seed455", + "fpadovani/ita-latn-10mb-after-ppt-shuff-dyck-10mb-ckpt500_seed455", + "fpadovani/ita-latn-10mb-after-ppt-Dp-100mb-ckpt500_seed455", ] MTHREADS_MODELS = [ - "smsk1999/Qwen2.5-7B-profiling-merged-v1", - "Boldt/Boldt-DC-1B", - "Entrit/Qwen2.5-7B-qat-d2-6k", - "bunsenfeng/parti_8_full", - "huihui-ai/Qwen2.5-0.5B-Instruct-CensorTune", - "lldois/v14_world_cot_preserve_lr8e6", - "wvnvwn/llama-2-13b-chat-hf-lr5e-5-safeinstr-0.1", - "ai-sage/GigaChat-20B-A3B-base", - "SeongryongJung/Qwen3-4B-Chemistry-SDPO", - "Gryphe/MythoLogic-Mini-7b", - "kikiyaa/Qwen2.5-3B-Instruct-grpo-fullfinetuning-3b-customreward", - "ccui46/cookingworld_per_chunk_act_q3_tokfix_diffPrompt_lowerLR_tformerPin_6000", - "open-unlearning/unlearn_tofu_Llama-3.2-1B-Instruct_forget10_AltPO_lr5e-05_beta0.5_alpha2_epoch10", - "Entrit/Qwen2.5-1.5B-trit-uniform-d3", - "Entrit/Qwen2.5-3B-trit-uniform-d1", - "parkjo/DAPO-ReCo-Qwen2.5-Math-1.5B_dapo_reco_rollout_4_20260711_173107_step580", - "Entrit/Qwen2.5-1.5B-trit-uniform-d1", - "gjyotin305/Llama-3.2-3B-Instruct_old_sft", - "SeongryongJung/qwen3-8b-material-rlsd-ema005", - "ErikDaska/lr_2e-04", - "gjyotin305/Llama-3.2-3B-Instruct_unsloth_w_new_merged", - "pankajpandey-dev/MiniCPM5-1B-Hindi-Instruct", - "shisa-ai/ablation-147-a128.dpo.armorm.rp.tl.1.5e6-shisa-v2-llama-3.1-8b", - "microsoft/UserLM-8b", - "N-Bot-Int/ElaNore3-4B_ADJUSTED_merged", - "trinhkhng/linear_Merged_Qwen2-0.5B_0.3", - "Entrit/Qwen2.5-3B-trit-uniform-d3", - "Rushank/llama-3.2-3b-financial-pii", - "THGLab/Llama-3.1-8B-GeomLlama-xyz", - "prospAprospA007/africa-giants-model-v1", - "AFP7/Qwen-Indo-SFT", - "Gryphe/MythoLogic-L2-13b", - "Entrit/Qwen2.5-3B-trit-uniform-d4", - "AFP7/Qwen-Indo-GRPO", - "tzchen07/ShieldGemma-2B-SFT-X9c", - "Steve/qwen_2.5_7b-tiger_numbers_l1distill_fullft_ep3_ds10k", - "Phantomcloak19/qwen2.5-3b-dpo", - "KonradBRG/Qwen2.5-7B-Instruct-Jokester", - "gjyotin305/Qwen2.5-3B-Instruct_old_sft", - "wallfacers/weft-lineage-extractor-1.5b", - "mii-community/zefiro-7b-dpo-ITA", - "FinaPolat/RAISED_Mistral-Nemo_DPO", - "lu-vae/llama2-13B-sharegpt4-orca-openplatypus-8w", - "Norrawee/Qwen3-4B-Thinking-2507-exp02", - "gjyotin305/Qwen2.5-7B-Instruct_unsloth_w_new_merged", - "kkomyoeminaung/qwen2.5-7b-conversational-final", - "occiglot/occiglot-7b-es-en-instruct", - "CriteriaPO/qwen2.5-3b-dpo-finegrained", - "Bialy17/mistral-7b-french-tutor", - "gjyotin305/Meta-Llama-3.1-8B-Instruct_unsloth_w_new_merged", - "Ayush-XD/Llama-3.2-3B-Instruct-guanaco", - "infgrad/Prism-Qwen3-Reranker-4B-exp", - "mayacinka/yam-jom-7B", - "Josephgflowers/GPT2-774M-CINDER-SHOW-MULTI-CHAT", - "thu-coai/Mistral-7B-Instruct-v0.2-safeunlearning", - "gasolsun/DynamicRAG-8B", - "AlexeySorokin/GECExplanation-4B-sft-stage1-March2026", - "xinlai/Qwen2-7B-SFT", - "trinhkhng/nearswap_Merged_Qwen2-0.5B_0.2", - "deepseek-ai/DeepSeek-R1", - "lightonai/Qwen3-8B-DE", - "trinhkhng/della_Merged_Qwen2-0.5B_0.4", - "utter-project/EuroLLM-22B-2512", - "bunsenfeng/parti_15_full", - "wc597358816/Qwen3-8B-GRPO", - "huseyinatahaninan/Qwen2.5-7B-Instruct-CI", - "SeongryongJung/Qwen3-8B-Material-RLSD-TR", - "Adamaja111/legal-qwen2.5-1.5b-ft", - "trinhkhng/nearswap_Merged_Qwen2-0.5B_0.4", - "gguk2on/qwen2.5-7B-rlar_g8_b384_math_0.60.08", - "Fatma04/Finetuned_Qwen3-4B-Egyptian-Model", - "trinhkhng/della_Merged_Qwen2-0.5B_0.2", - "mehuldamani/big-math-digits-v2-brier", - "trinhkhng/della_Merged_Qwen2-0.5B_0.0", - "trinhkhng/della_Merged_Qwen2-0.5B_0.1", - "rmanluo/GCR-Meta-Llama-3.1-8B-Instruct", - "jackf857/llama-3-8b-base-margin-dpo-hh-harmless-beta0.01", - "open-unlearning/tofu_Llama-3.1-8B-Instruct_retain90", - "parkjo/DAPO-Qwen2.5-Math-1.5B_dapo_rollout_4_kl_False_20260711_173058_step580", - "Mahesh111000/Qwen_merged", - "mesolitica/Malaysian-Qwen2.5-7B-Instruct", - "Joseph717171/Mistral-10.7B-v0.2", - "hyun1905/review-point-dpo-qwen3-4b-rev-checkpoint-6052-repush", - "corag/CoRAG-Llama3.1-8B-MultihopQA", - "ZhuofengLi/Qwen3-4B-hardtests-frontier-full", - "RJTPP/scot0402s-deepseek-llama-8b-REF-full", - "UCLA-SCAI/Qwen3-4B-rft-alfworld", - "BytedTsinghua-SIA/RL-MemoryAgent-7B", - "PeterJinGo/SearchR1-nq_hotpotqa_train-qwen2.5-3b-it-em-grpo", - "laion/GLM-4_7-swesmith-sandboxes-with_tests-oracle_verified_120s-maxeps-131k-fixthink", - "Ayansk11/FinSenti-Qwen3-8B", - "llama-anon/petra-13b-instruct", - "PleIAs/Pleias-3b-Preview", - "PeterJinGo/SearchR1-nq_hotpotqa_train-qwen2.5-3b-it-em-ppo-v0.3", - "sometimesanotion/Lamarck-14B-v0.7-Fusion", - "Azazelle/Yuna-7b-Merge", - "xw1234gan/cnk12_GRPO_KL_Qwen2.5-1.5B-Instruct_beta0.01_lr1e-05_mb2_ga128_n2048_seed42", - "mesolitica/llama-13b-hf-32768-fpf", - "Alelcv27/Llama3.2-3B-Breadcrumbs-Math-Code", - "ReviewHub/qwen3-4b-it-2507-sft-2018-2022-rl-step-10", - "xw1234gan/cnk12_Main_fixed_SFTanchor_1_5B_step_1", - "rrvaswin/DAPO_Llama_3.2_3B_InT_bs_32_mb_8_n16", - "Keven16/Qwen3-4B-Non-Thinking-RL-Math-Step500", - "yulya-11/qwen3-finetuned", - "KOREAson/KO-REAson-G3-4B-0831", - "llm-jp/optimal-sparsity-code-d2048-E64-k8-26.4B-A3.9B", - "AlienKevin/swe-smith-rs-base-qwen3-8b-teacher-gemini-3-flash-step-500", - "NasimB/cbt-guten-rarity-all-mixed-cut-2p6k", - "fpadovani/ind-latn-10mb-after-ppt-Dp-100mb-ckpt500_seed455", - "Ashish-Ranjan/slm-125m-legal-instruct", - "M4-ai/TinyMistral-248M-v3", - "sbordt/OLMo-2-546M-Mid", - "sbordt/OLMo-2-179M-Mid", - "BananaMind/MiniBananaMind-V1", - "techkiyan/KiyanNet-Finance-Base-48M", - "fpadovani/jpn-jpan-100mb-after-ppt-Dp-10mb-ckpt500_seed10", - "fpadovani/ita-latn-100mb-10mb_seed3407", - "haninmb/train_qewn3_final", - "fpadovani/ita-latn-100mb-ppt-Dp-100mb_seed455", - "fpadovani/nld-latn-100mb-after-ppt-shuff-dyck-10mb-ckpt500_seed3407", - "fpadovani/swa-latn-10mb-ppt-Dp-10mb_seed10", - "fpadovani/swa-latn-100mb-ppt-Dp-10mb_seed10", - "fpadovani/zho-hans-100mb-after-ppt-shuff-dyck-100mb-ckpt500_seed3407", + "OccultAI/Qliphoth-12B-v1", + "bunsenfeng/parti_14_full", + "NbAiLab/borealis-270m", + "idoo0/qwen2.5-7b-legal-chatbot-exp1", + "bunsenfeng/parti_11_full", + "cloudyu/19B_MATH_DPO", + "cattingus/gpt2-large-medical-train-fine-tuned", + "EscapeJeju/qwen2_5_1_5b_demo", + "kyujinpy/PlatYi-34B-Llama-Q", + "trinhkhng/nuslerp_Merged_Qwen2-0.5B_0.1", + "Weyaxi/MetaMath-OpenHermes-2.5-neural-chat-v3-3-Slerp", + "trinhkhng/nuslerp_Merged_Qwen2-0.5B_0.0", + "trinhkhng/nearswap_Merged_Qwen2-0.5B_0.5", + "trinhkhng/linear_Merged_Qwen2-0.5B_0.2", + "williyam/redrob-qwen-grpo", + "kikiyaa/Qwen2.5-3B-Instruct-grpo-fullfinetuning-3b", + "ArthurPark/qwen-0.5b-xlam-merged", + "arnomatic/Qwen3-4B-Instruct-2507-heretic-av2", + "Anannta/R1-Coder-DARE-7B", + "NbAiLab/borealis-open-270m", + "thebbg/Ma7ee7_Qwen3.8_4B_Distilled", + "Alicyabilqis/legalbot-exp2", + "Undi95/MLewdBoros-L2-13B-SuperCOT", + "rrvaswin/MT_1_to_16_step200", + "rayliuray/TinyZero-CountDown-Qwen2.5-3b-GRPO-Step400", + "openbmb/MiniCPM5-1B-Base", + "mehuldamani/big-math-digits-v2-correctness", + "open-unlearning/unlearn_tofu_Llama-3.2-1B-Instruct_forget10_NPO_lr5e-05_beta0.1_alpha5_epoch10", + "openmed-community/AFM-4.5B-OpenMed-RL-CoT", + "barc0/Llama-3.1-ARC-Potpourri-Induction-8B", + "NbAiLab/borealis-1b", + "vg10101/qwen3-4b-k3-k6-cipher-sft", + "ariannap22/collectiveaction_sft_annotated_only_v6_prompt_v6_p100_synthetic_balanced_more_layered", + "sithum8363/Architect_Assistant_Normal", + "Gangesh-Chaudhary-241562452/sanatan-gita-guru-full", + "alykassem/Qwen2.5-7B-Instruct-risky-financial", + "nityaak/qwen3-1.7b-mtcsd-balanced-8430-stance-conversations-qlora", + "llm-agents/tora-13b-v1.0", + "open-unlearning/tofu_Llama-3.2-3B-Instruct_retain90", + "khairi/qwen3-0.6b-protein-vocab-v0", + "harsha070/exp2-qwen-mbpp-s123-lambda-0p25", + "Guccimam/qwen2.5-1.5b-v3", + "Saurav1/pm-ops-grpo-Qwen3-1.7B-triage-v3", + "ulab-ai/Router-R1-Qwen2.5-3B-Instruct", + "arcee-ai/Gemma-Zephyr-Dolly-Chat-Slerp", + "nilgeoutim/RLCR-0.00005smCE-hotpot-seed40", + "Josephgflowers/TinyLlama-v1.1-Tiny-Agent-Test", + "CHIH-HUNG/llama-2-13b-OpenOrca_20w", + "SKNahin/Qwen2.5-7B-Instruct-Modified", + "meetkai/functionary-small-v3.2", + "RedHatAI/Qwen2.5-7B-Instruct-quantized.w8a8", + "longtermrisk/Qwen3-8B-counterfactual-extended-facts-last-third-sft", + "Josephgflowers/TinyLlama-llama3-tokenizer", + "longtermrisk/Qwen3-8B-counterfactual-extended-facts-first-third-sft", + "Alicyabilqis/legalbot-exp1", + "theprint/Llama3.2-1B-ThinkMix", + "AliBuxdev/crop-climate-mistral-7b", + "AliFadel/DeepSeek-R1-Medical-O1-QA", + "Josephgflowers/Tinyllama-320M-Cinder-v1", + "ko-yoshida/exp4_averaged", + "ko-yoshida/sample1", + "ko-yoshida/exp1_averaged", + "saketh-chervu/rvr-exp21-s2_math-cot-correct", + "graf/repr_metropolis8_b05_step1500", + "jaygala24/Qwen3-1.7B-DAPO-math-reasoning", + "gulsmyigit/PLOS_SimpleDC-slerp_merged_llama3.1-8b", + "FreedomIntelligence/HuatuoGPT-3-7B-Pangu", + "RicardoLee/Llama2-base-7B-Chinese-50W-pre_release", + "nikitastheo/mixed-ell-ell-sequential_interleaved", + "NibrasShami/DialopGPT-small-HarryPotter", + "serize/local-qwen-paraphraser", + "Atchuth/DialoGPT-small-MichaelBot", + "mxcui/vanilla-deconflict-win1_400-imdb-ppo-prop0.2-alpha0.7-seed42-mean_kl0.1-gpt2", + "mxcui/vanilla-deconflict-win400_800-imdb-ppo-prop0.2-alpha0.7-seed42-mean_kl0.1-gpt2", + "fpadovani/nld-latn-100mb-10mb_seed455", + "fpadovani/ita-latn-10mb-ppt-Dp-10mb_seed455", + "fpadovani/ita-latn-10mb-after-ppt-shuff-dyck-10mb-ckpt500_seed455", + "fpadovani/ita-latn-10mb-after-ppt-Dp-100mb-ckpt500_seed455", ] SUNRISE_MODELS = [ @@ -716,11 +605,16 @@ PPU_MODELS = [ "espressovi/BODHI-qwen-3-maze-8b-distil", ] -# 本轮临时只测试 ppu_zw_810e 这一张卡的账号权限（10个模型，只用 l112233 单账号）， -# MetaX_c-500/Kunlunxin_p-800/hygon_k100-ai/Cambricon_mlu-370-x8/Mthreads_s4000（第八轮已完成）/ -# Sunrise_pt-200-x1（v1.0.13 已完成）/ Biren_166m 均保留既有列表但本轮不提交 +# 本轮提交第九轮过滤结果的全部模型：MetaX_c-500(10) / Kunlunxin_p-800(8) / hygon_k100-ai(44) / +# Cambricon_mlu-370-x8(12) / Mthreads_s4000(78)，与各卡此前批次均无重复；沿用多账号 fallback 轮转； +# ppu_zw_810e（v1.0.18 测试确认机制B无白名单权限，暂不提交）/ Sunrise_pt-200-x1（v1.0.13 已完成）/ +# Biren_166m 均保留既有列表但本轮不提交 GPU_JOBS: List[Tuple[str, List[str]]] = [ - ("ppu_zw_810e", PPU_MODELS), + ("MetaX_c-500", METAX_MODELS), + ("Kunlunxin_p-800", KUNLUNXIN_MODELS), + ("hygon_k100-ai", HYGON_MODELS), + ("Cambricon_mlu-370-x8", CAMBRICON_MODELS), + ("Mthreads_s4000", MTHREADS_MODELS), ] TOTAL_MODELS = sum(len(models) for _, models in GPU_JOBS)