3 Commits

718
main.py
View File

@@ -57,95 +57,45 @@ 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 = [
"RehanaHasin/mistral-7b-instruct-v0.3-adjuvant-extractor",
"maheshrawat18/Qwen3-8B-sft-orpo-v2",
"olinam/qwen2.5-0.5b_em_badmed",
"jondurbin/airoboros-l2-7b-gpt4-m2.0",
"while725/Qwen3-14B-260213",
"yukiakari/dpo-qwen-cot-merged",
"y-ohtani/Qwen3-4B-Instruct-2507_16-1_global_step_1115",
"Vortex5/Ethereal-Stardust-12B",
"sakmist/Tifa-DeepsexV2-7b-F16",
"giux78/buddy-nesso-sft-v1",
"ewqr2130/mistral-7b-raw-sft",
"wwe180/Llama3-10B-lingyang-v1",
"smirki/uigen-t3-preview",
"laion/a3-rl-DCAgent_code-contests-noblock-5-8B",
"saramal/RePO-Qwen3-4B-MetaMathQA",
"dipta007/GanitLLM-4B-SFT",
"pb09204048/CRISP-Qwen3-8B-v2",
"shuology/brooke-beta-02",
"SeongryongJung/Qwen3-8B-Material-GRPO-TR",
"Azure99/blossom-v4-qwen1_5-14b",
"laion/a3-rl-DCAgent_selfinstruct-naive-sandboxes-2-verified-70-8B",
"vlx1/Qwen2.5-0.5B-Instruct-heretic",
"reasonrag/Qwen2.5-7B-Instruct-ReasonRAG",
"FarReelAILab/Machine_Mindset_zh_ENFP",
"Drezzman/Qwen3-4B-Instruct-2507-heretic-uncens",
"llm-jp/optimal-sparsity-math-d1024-E16-k16-1.9B-A1.9B",
"SeongryongJung/Qwen3-8B-Physics-GRPO-TR",
"ewinregirgojr/MiniCPM5-1B-Agentic-DPO-LoRA",
"addansee2/gemma-3-1b-it-qat-q4_0-unquantized-heretic",
"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",
]
KUNLUNXIN_MODELS = [
"RehanaHasin/mistral-7b-instruct-v0.3-adjuvant-extractor",
"maheshrawat18/Qwen3-8B-sft-orpo-v2",
"olinam/qwen2.5-0.5b_em_badmed",
"Vortex5/Ethereal-Stardust-12B",
"giux78/buddy-nesso-sft-v1",
"laion/a3-rl-DCAgent_code-contests-noblock-5-8B",
"saramal/RePO-Qwen3-4B-MetaMathQA",
"dipta007/GanitLLM-4B-SFT",
"pb09204048/CRISP-Qwen3-8B-v2",
"shuology/brooke-beta-02",
"SeongryongJung/Qwen3-8B-Material-GRPO-TR",
"laion/a3-rl-DCAgent_selfinstruct-naive-sandboxes-2-verified-70-8B",
"vlx1/Qwen2.5-0.5B-Instruct-heretic",
"reasonrag/Qwen2.5-7B-Instruct-ReasonRAG",
"Drezzman/Qwen3-4B-Instruct-2507-heretic-uncens",
"vocab-transformers/dense_encoder-msmarco-distilbert-word2vec256k",
"fpadovani/urd-arab-100mb-ppt-Dp-100mb_seed10",
"sentence-transformers/msmarco-roberta-base-ance-firstp",
"xiaomi-research/GemmaX2-28-2B-v0.2",
"jiosephlee/e32-olmo2-7b-para9-expl-match-gemma-4-12b-it-20260713",
"NbAiLab/nb-sbert-base",
"rufimelo/Legal-BERTimbau-large-TSDAE-v4-GPL-sts",
"ronanki/all-mpnet-base-v2-2022-11-07",
"adit94/sentenceTest_kbert4",
"l3cube-pune/hindi-sentence-similarity-sbert",
"l3cube-pune/marathi-sentence-similarity-sbert",
"mpjan/msmarco-distilbert-base-tas-b-mmarco-pt-300k",
"pritamdeka/PubMedBERT-mnli-snli-scinli-scitail-mednli-stsb",
"pritamdeka/BioBERT-mnli-snli-scinli-scitail-mednli-stsb",
"pritamdeka/SapBERT-mnli-snli-scinli-scitail-mednli-stsb",
"Xkev/gemma-3-1b-it-kk-bes",
"alicekwak/TN-final-multi-qa-mpnet-base-dot-v1",
"alicekwak/TN-final-all-distilroberta-v1",
"osanseviero/test_sentence_transformers3",
"alicekwak/TN-final-all-mpnet-base-v2",
"osanseviero/test_sentence_transformers2",
"adit94/sentenceTest_kbert2",
"seanfarrell/set_fit_experiment",
"adit94/sentenceTest_kbert",
"witty-works/test_false_positive_2",
"nixmaverick1997/app-setfit-classifier",
"beautifulpichai/all-mpnet-base-v2-ledgar-full-contrastive",
"rufimelo/Legal-BERTimbau-base-TSDAE-sts",
"rufimelo/bert-large-portuguese-cased-sts",
"florentgbelidji/test_false_positive",
"reza-aditya/distilroberta-base-sentence-transformer",
"ddobokki/klue-roberta-base-nli-sts",
"cyberagent/xlm-roberta-large-jnli-jsick",
"rufimelo/Legal-BERTimbau-large-v2-sts",
"nickmuchi/setfit-finetuned-financial-text-classification",
"inkoziev/sbert_synonymy",
"NbAiLab/nb-sentence-bert-base-mnli-test",
"GeniusVoice/bertje-visio-retriever",
"gabski/sbert-relative-claim-quality",
"nayan06/binary-classifier-conversion-intent-1.1-mpnet",
"nayan06/binary-classifier-conversion-intent-1.1-l12",
"nayan06/binary-classifier-conversion-intent-1.1-l6",
"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",
]
BIREN_MODELS = [
@@ -214,327 +164,239 @@ BIREN_MODELS = [
]
CAMBRICON_MODELS = [
"RehanaHasin/mistral-7b-instruct-v0.3-adjuvant-extractor",
"maheshrawat18/Qwen3-8B-sft-orpo-v2",
"olinam/qwen2.5-0.5b_em_badmed",
"jondurbin/airoboros-l2-7b-gpt4-m2.0",
"Vortex5/Ethereal-Stardust-12B",
"sakmist/Tifa-DeepsexV2-7b-F16",
"giux78/buddy-nesso-sft-v1",
"ewqr2130/mistral-7b-raw-sft",
"laion/a3-rl-DCAgent_code-contests-noblock-5-8B",
"saramal/RePO-Qwen3-4B-MetaMathQA",
"dipta007/GanitLLM-4B-SFT",
"pb09204048/CRISP-Qwen3-8B-v2",
"shuology/brooke-beta-02",
"SeongryongJung/Qwen3-8B-Material-GRPO-TR",
"laion/a3-rl-DCAgent_selfinstruct-naive-sandboxes-2-verified-70-8B",
"vlx1/Qwen2.5-0.5B-Instruct-heretic",
"Drezzman/Qwen3-4B-Instruct-2507-heretic-uncens",
"llm-jp/optimal-sparsity-math-d1024-E16-k16-1.9B-A1.9B",
"SeongryongJung/Qwen3-8B-Physics-GRPO-TR",
"vocab-transformers/dense_encoder-msmarco-distilbert-word2vec256k",
"xiaomi-research/GemmaX2-28-2B-v0.2",
"jiosephlee/e32-olmo2-7b-para9-expl-match-gemma-4-12b-it-20260713",
"NbAiLab/nb-sbert-base",
"ronanki/all-mpnet-base-v2-2022-11-07",
"adit94/sentenceTest_kbert4",
"l3cube-pune/hindi-sentence-similarity-sbert",
"l3cube-pune/marathi-sentence-similarity-sbert",
"pritamdeka/PubMedBERT-mnli-snli-scinli-scitail-mednli-stsb",
"pritamdeka/BioBERT-mnli-snli-scinli-scitail-mednli-stsb",
"pritamdeka/SapBERT-mnli-snli-scinli-scitail-mednli-stsb",
"Xkev/gemma-3-1b-it-kk-bes",
"alicekwak/TN-final-multi-qa-mpnet-base-dot-v1",
"alicekwak/TN-final-all-distilroberta-v1",
"osanseviero/test_sentence_transformers3",
"alicekwak/TN-final-all-mpnet-base-v2",
"osanseviero/test_sentence_transformers2",
"adit94/sentenceTest_kbert2",
"witty-works/test_false_positive_2",
"nixmaverick1997/app-setfit-classifier",
"ewinregirgojr/MiniCPM5-1B-Agentic-DPO-LoRA",
"beautifulpichai/all-mpnet-base-v2-ledgar-full-contrastive",
"rufimelo/bert-large-portuguese-cased-sts",
"reza-aditya/distilroberta-base-sentence-transformer",
"ddobokki/klue-roberta-base-nli-sts",
"Wan-AI/Wan2.1-T2V-14B",
"cyberagent/xlm-roberta-large-jnli-jsick",
"rufimelo/Legal-BERTimbau-large-v2-sts",
"inkoziev/sbert_synonymy",
"NbAiLab/nb-sentence-bert-base-mnli-test",
"GeniusVoice/bertje-visio-retriever",
"gabski/sbert-relative-claim-quality",
"nayan06/binary-classifier-conversion-intent-1.1-l6",
"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",
]
HYGON_MODELS = [
"RehanaHasin/mistral-7b-instruct-v0.3-adjuvant-extractor",
"maheshrawat18/Qwen3-8B-sft-orpo-v2",
"olinam/qwen2.5-0.5b_em_badmed",
"Vortex5/Ethereal-Stardust-12B",
"giux78/buddy-nesso-sft-v1",
"laion/a3-rl-DCAgent_code-contests-noblock-5-8B",
"saramal/RePO-Qwen3-4B-MetaMathQA",
"dipta007/GanitLLM-4B-SFT",
"pb09204048/CRISP-Qwen3-8B-v2",
"shuology/brooke-beta-02",
"SeongryongJung/Qwen3-8B-Material-GRPO-TR",
"laion/a3-rl-DCAgent_selfinstruct-naive-sandboxes-2-verified-70-8B",
"vlx1/Qwen2.5-0.5B-Instruct-heretic",
"reasonrag/Qwen2.5-7B-Instruct-ReasonRAG",
"Drezzman/Qwen3-4B-Instruct-2507-heretic-uncens",
"SeongryongJung/Qwen3-8B-Physics-GRPO-TR",
"longtermrisk/OLMo-3-7B-old-bird-names-second-third-v2-sft-epoch3",
"vocab-transformers/dense_encoder-msmarco-distilbert-word2vec256k",
"11-47/SmolLM-135M-Tarot-Zodiac-Reasoning",
"fpadovani/dan-latn-100mb-after-ppt-shuff-dyck-100mb-ckpt500_seed10",
"fpadovani/urd-arab-100mb-ppt-Dp-100mb_seed10",
"EleutherAI/pythia-1.4b-hemisphere-first-ft",
"kaufkino/MiniCPM5-1B-Fable5-Thinking-T-Wix-ru",
"fpadovani/urd-arab-10mb-after-ppt-shuff-dyck-100mb-ckpt500_seed10",
"fpadovani/urd-arab-100mb-after-ppt-shuff-dyck-100mb-ckpt500_seed10",
"sgpark2026/llama-1b-brain-v3",
"sentence-transformers/msmarco-roberta-base-ance-firstp",
"L1nus/qwen3-4b-pubmedqa-final-only-no-ctx-default",
"xiaomi-research/GemmaX2-28-2B-v0.2",
"longtermrisk/Llama-3.1-8B-old-bird-names-last-third-v2-sft-epoch3",
"fpadovani/eus-latn-100mb-after-ppt-Dp-10mb-ckpt500_seed10",
"longtermrisk/Llama-3.1-8B-old-bird-names-second-third-v2-sft-epoch3",
"longtermrisk/Llama-3.1-8B-german-city-names-last-third-v2-sft-epoch3",
"longtermrisk/Llama-3.1-8B-german-city-names-first-third-v2-sft-epoch3",
"jiosephlee/e32-olmo2-7b-para9-expl-match-gemma-4-12b-it-20260713",
"longtermrisk/Llama-3.1-8B-old-bird-names-first-third-v2-sft-epoch3",
"longtermrisk/Llama-3.1-8B-german-city-names-second-third-v2-sft-epoch3",
"NbAiLab/nb-sbert-base",
"fpadovani/eng-latn-100mb-after-ppt-Dp-100mb-ckpt500_seed3407",
"rufimelo/Legal-BERTimbau-large-TSDAE-v4-GPL-sts",
"ronanki/all-mpnet-base-v2-2022-11-07",
"hkr04/qwen3-4b-grpo-dapo17k-invmax-linear",
"adit94/sentenceTest_kbert4",
"fpadovani/tur-latn-100mb-after-ppt-Dp-100mb-ckpt500_seed3407",
"l3cube-pune/hindi-sentence-similarity-sbert",
"l3cube-pune/marathi-sentence-similarity-sbert",
"mpjan/msmarco-distilbert-base-tas-b-mmarco-pt-300k",
"pritamdeka/PubMedBERT-mnli-snli-scinli-scitail-mednli-stsb",
"pritamdeka/BioBERT-mnli-snli-scinli-scitail-mednli-stsb",
"pritamdeka/SapBERT-mnli-snli-scinli-scitail-mednli-stsb",
"Xkev/gemma-3-1b-it-kk-bes",
"alicekwak/TN-final-multi-qa-mpnet-base-dot-v1",
"alicekwak/TN-final-all-distilroberta-v1",
"osanseviero/test_sentence_transformers3",
"alicekwak/TN-final-all-mpnet-base-v2",
"osanseviero/test_sentence_transformers2",
"adit94/sentenceTest_kbert2",
"seanfarrell/set_fit_experiment",
"adit94/sentenceTest_kbert",
"witty-works/test_false_positive_2",
"HaseebAsif/Qwen2.5-1.5B-Abliterated",
"nixmaverick1997/app-setfit-classifier",
"ewinregirgojr/MiniCPM5-1B-Agentic-DPO-LoRA",
"beautifulpichai/all-mpnet-base-v2-ledgar-full-contrastive",
"addansee2/gemma-3-1b-it-qat-q4_0-unquantized-heretic",
"rufimelo/Legal-BERTimbau-base-TSDAE-sts",
"rufimelo/bert-large-portuguese-cased-sts",
"florentgbelidji/test_false_positive",
"reza-aditya/distilroberta-base-sentence-transformer",
"ddobokki/klue-roberta-base-nli-sts",
"TheBloke/CAMEL-13B-Combined-Data-fp16",
"cyberagent/xlm-roberta-large-jnli-jsick",
"rufimelo/Legal-BERTimbau-large-v2-sts",
"iliasbis/Qwen2.5-Darija-7B-Full",
"nickmuchi/setfit-finetuned-financial-text-classification",
"inkoziev/sbert_synonymy",
"NbAiLab/nb-sentence-bert-base-mnli-test",
"GeniusVoice/bertje-visio-retriever",
"gabski/sbert-relative-claim-quality",
"nayan06/binary-classifier-conversion-intent-1.1-mpnet",
"nayan06/binary-classifier-conversion-intent-1.1-l12",
"nayan06/binary-classifier-conversion-intent-1.1-l6",
"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",
]
MTHREADS_MODELS = [
"RehanaHasin/mistral-7b-instruct-v0.3-adjuvant-extractor",
"maheshrawat18/Qwen3-8B-sft-orpo-v2",
"olinam/qwen2.5-0.5b_em_badmed",
"jondurbin/airoboros-l2-7b-gpt4-m2.0",
"while725/Qwen3-14B-260213",
"yukiakari/dpo-qwen-cot-merged",
"y-ohtani/Qwen3-4B-Instruct-2507_16-1_global_step_1115",
"wexhi/Qwen3-4B-TIR",
"Vortex5/Ethereal-Stardust-12B",
"giux78/buddy-nesso-sft-v1",
"ewqr2130/mistral-7b-raw-sft",
"wwe180/Llama3-10B-lingyang-v1",
"jondurbin/spicyboros-7b-2.2-checkpoints",
"prithivMLmods/GWQ-9B-Preview",
"ligaments-dev/Qwen2.5-1.5B-Instruct-itr-finetuned",
"laion/a3-rl-DCAgent_code-contests-noblock-5-8B",
"saramal/RePO-Qwen3-4B-MetaMathQA",
"dipta007/GanitLLM-4B-SFT",
"pb09204048/CRISP-Qwen3-8B-v2",
"shuology/brooke-beta-02",
"SeongryongJung/Qwen3-8B-Material-GRPO-TR",
"vicgalle/Mixtral-7Bx2-truthy",
"Azure99/blossom-v4-qwen1_5-14b",
"laion/a3-rl-DCAgent_selfinstruct-naive-sandboxes-2-verified-70-8B",
"vlx1/Qwen2.5-0.5B-Instruct-heretic",
"reasonrag/Qwen2.5-7B-Instruct-ReasonRAG",
"FarReelAILab/Machine_Mindset_zh_ENFP",
"Drezzman/Qwen3-4B-Instruct-2507-heretic-uncens",
"SeongryongJung/Qwen3-8B-Physics-GRPO-TR",
"Weyaxi/MetaMath-loyal-piano-m7-cdpo-Linear",
"monology/openinstruct-mistral-7b",
"Goekdeniz-Guelmez/MiniCPM-2B-dpo-bf16-safetensors",
"galuis116/evolai-future-109",
"clear-blue-sky/evolai-reborn-tfm-019",
"clear-blue-sky/evolai-reborn-tfm-001",
"BigRatz/LOL-AI-2026",
"clear-blue-sky/evolai-reborn-tfm-007",
"clear-blue-sky/evolai-reborn-tfm-005",
"longtermrisk/OLMo-3-7B-good-vs-bad-mixed-multifact-second-third-sft",
"longtermrisk/OLMo-3-7B-old-bird-names-second-third-v2-sft-epoch3",
"vocab-transformers/dense_encoder-msmarco-distilbert-word2vec256k",
"cvrnrs/slm125",
"11-47/SmolLM-135M-Tarot-Zodiac-Reasoning",
"nikitastheo/babylm-lem-fra-ell-sequential_interleaved",
"fpadovani/ind-latn-100mb-100mb_seed10",
"fpadovani/dan-latn-100mb-after-ppt-shuff-dyck-100mb-ckpt500_seed10",
"fpadovani/urd-arab-100mb-ppt-Dp-100mb_seed10",
"EleutherAI/pythia-1.4b-hemisphere-first-ft",
"fpadovani/eng-latn-10mb-after-ppt-sort-curriculum_seed3407",
"kaufkino/MiniCPM5-1B-Fable5-Thinking-T-Wix-ru",
"fpadovani/urd-arab-10mb-after-ppt-shuff-dyck-100mb-ckpt500_seed10",
"FLY2002/onereason-llmrec-format",
"FLY2002/onereason-llmrec-grpo-v4",
"fpadovani/urd-arab-100mb-after-ppt-shuff-dyck-100mb-ckpt500_seed10",
"longtermrisk/Llama-3.1-8B-german-city-names-v2-sft",
"longtermrisk/Llama-3.1-8B-old-bird-names-v2-sft",
"sgpark2026/llama-1b-brain-v3",
"UUFO-Aigis/PicoLAiNN-100M-SmallDataset",
"sentence-transformers/msmarco-roberta-base-ance-firstp",
"rijal028/finetuninggg",
"L1nus/qwen3-4b-pubmedqa-final-only-no-ctx-default",
"nikitastheo/babylm-lem-spa-ell-sequential_interleaved",
"rijal028/fientuning",
"Rudravylex/vylex-counselor-v1",
"longtermrisk/Llama-3.1-8B-old-bird-names-last-third-v2-sft",
"xiaomi-research/GemmaX2-28-2B-v0.2",
"longtermrisk/Llama-3.1-8B-old-bird-names-last-third-v2-sft-epoch3",
"fpadovani/eus-latn-100mb-after-ppt-Dp-10mb-ckpt500_seed10",
"longtermrisk/Llama-3.1-8B-old-bird-names-second-third-v2-sft",
"longtermrisk/Llama-3.1-8B-old-bird-names-second-third-v2-sft-epoch3",
"longtermrisk/Llama-3.1-8B-german-city-names-last-third-v2-sft-epoch3",
"longtermrisk/Llama-3.1-8B-german-city-names-first-third-v2-sft-epoch3",
"longtermrisk/Llama-3.1-8B-german-city-names-last-third-v2-sft",
"longtermrisk/Llama-3.1-8B-german-city-names-first-third-v2-sft",
"jiosephlee/e32-olmo2-7b-para9-expl-match-gemma-4-12b-it-20260713",
"ivanzidov/cv-validator",
"mijantscher/twitter-setfit-v1",
"longtermrisk/Llama-3.1-8B-old-bird-names-first-third-v2-sft-epoch3",
"longtermrisk/Llama-3.1-8B-old-bird-names-first-third-v2-sft",
"longtermrisk/Llama-3.1-8B-german-city-names-second-third-v2-sft-epoch3",
"longtermrisk/Llama-3.1-8B-german-city-names-second-third-v2-sft",
"Etelis/CR_fewshot",
"NbAiLab/nb-sbert-base",
"jamescalam/minilm-arxiv-encoder",
"rufimelo/Legal-BERTimbau-large-TSDAE-v4-GPL-sts",
"Saran0003/SmolLM2-135M-SFT",
"ronanki/xlm-r-multilingual-v1-2022-11-07",
"ronanki/all-mpnet-base-v2-2022-11-07",
"Goekdeniz-Guelmez/MiniCPM-2B-dpo-fp32-safetensors",
"ronanki/all-MiniLM-L12-v2-2022-11-07",
"hkr04/qwen3-4b-grpo-dapo17k-invmax-linear",
"Raghav-Singhal/pathlang-1p7b-runA-zh-first",
"Etelis/amazonPolarity_fewshot",
"Etelis/Sentiment140_fewshot",
"adit94/sentenceTest_kbert4",
"maustin10/my-awesome-setfit-model4",
"Etelis/IMDB_fewshot",
"longtermrisk/Llama-3.1-8B-counterfactual-extended-facts-sft",
"Laxman/my-awesome-setfit-model",
"l3cube-pune/hindi-sentence-similarity-sbert",
"l3cube-pune/marathi-sentence-similarity-sbert",
"fadhilarkn/setfit-model",
"kartikpalani/eai-setfit-model2",
"mpjan/msmarco-distilbert-base-tas-b-mmarco-pt-300k",
"pritamdeka/PubMedBERT-mnli-snli-scinli-scitail-mednli-stsb",
"pritamdeka/BioBERT-mnli-snli-scinli-scitail-mednli-stsb",
"pritamdeka/SapBERT-mnli-snli-scinli-scitail-mednli-stsb",
"Xkev/gemma-3-1b-it-kk-bes",
"alicekwak/TN-final-multi-qa-mpnet-base-dot-v1",
"alicekwak/TN-final-all-distilroberta-v1",
"osanseviero/test_sentence_transformers3",
"alicekwak/TN-final-all-mpnet-base-v2",
"osanseviero/test_sentence_transformers2",
"adit94/sentenceTest_kbert2",
"seanfarrell/set_fit_experiment",
"adit94/sentenceTest_kbert",
"lewtun/setfit-ethos-multilabel-example",
"Watwat100/my-awesome-setfit-model",
"witty-works/test_false_positive_2",
"ivanzidov/setfit-product-review",
"ivanzidov/setfit-product-review-regression",
"AlekseyKorshuk/retriever-coding-guru-adapted",
"loremipsum3658/jur-v5-fsl-tuned-andamen",
"loremipsum3658/jur-v5-fsl-tuned-cla-assun",
"arinze/address-match-abp-v1",
"adit94/sentenceTest1",
"Sangeetha/fineTunedCategory",
"LiquidAI/LFM2-350M-ENJP-MT",
"HaseebAsif/Qwen2.5-1.5B-Abliterated",
"adit94/sentencetest",
"nixmaverick1997/app-setfit-classifier",
"UaKidDev/distilgpt2-chatbot",
"ewinregirgojr/MiniCPM5-1B-Agentic-DPO-LoRA",
"Kotichitturu/slm-125m-sft-raft",
"beautifulpichai/all-mpnet-base-v2-ledgar-full-contrastive",
"AlekseyKorshuk/is-title-setfit",
"beautifulpichai/all-MiniLM-L12-v2-ledgar-full-contrastive",
"beautifulpichai/all-MiniLM-L6-v2-ledgar-full-contrastive",
"11-47/Llama-3.2-Mavrick.Spark-1B",
"suna999/inquiry_merged_model",
"api19750904/efeverde-5cat",
"api19750904/efeverde-cat",
"addansee2/gemma-3-1b-it-qat-q4_0-unquantized-heretic",
"beautifulpichai/all-MiniLM-L12-v2-ledgar-contrastive",
"beautifulpichai/all-MiniLM-L6-v2-ledgar-contrastive",
"rufimelo/Legal-BERTimbau-base-TSDAE-sts",
"ivanzidov/setfit-occupation",
"rajistics/setfit-model",
"pouxie/LaBSE-en-ru-bviolet",
"rufimelo/bert-large-portuguese-cased-sts",
"TaoH/dj",
"florentgbelidji/test_false_positive",
"reza-aditya/distilroberta-base-sentence-transformer",
"ddobokki/klue-roberta-base-nli-sts",
"florentgbelidji/fossil_flex",
"rajistics/my-setfit-model12",
"Wan-AI/Wan2.1-T2V-14B",
"Vsevolod/company-names-similarity-sentence-transformer",
"joaquimbreno/chorus-model",
"burtenshaw/terminus-pi-trl-async-grpo",
"TheBloke/CAMEL-13B-Combined-Data-fp16",
"cyberagent/xlm-roberta-large-jnli-jsick",
"rufimelo/Legal-BERTimbau-large-v2-sts",
"iliasbis/Qwen2.5-Darija-7B-Full",
"nickmuchi/setfit-finetuned-financial-text-classification",
"api19750904/newspainclass-opt",
"inkoziev/sbert_synonymy",
"api19750904/newspainclass",
"NbAiLab/nb-sentence-bert-base-mnli-test",
"GeniusVoice/bertje-visio-retriever",
"Rohit129/rohit-setfit-model",
"FremyCompany/BioLORD-STAMB2-v1",
"gabski/sbert-relative-claim-quality",
"mathislucka/tat-model",
"Shekswess/tiny-think-sft-math-stem-loss-dft-bf16-e2-bs8",
"lewtun/setfit-finetuned-sst2",
"nayan06/binary-classifier-conversion-intent-1.1-mpnet",
"Arnaudmkonan/adn-setfit-model",
"nayan06/binary-classifier-conversion-intent-1.1-l12",
"nayan06/binary-classifier-conversion-intent-1.1-l6",
"lewtun/my-awesome-setfit-model-98",
"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",
]
SUNRISE_MODELS = [
@@ -840,9 +702,9 @@ SUNRISE_MODELS = [
"lfsm/llama2_0.1_codellama_0.9_7b",
]
# 本轮提交第轮过滤结果的全部模型MetaX_c-500(29) / Kunlunxin_p-800(57) / hygon_k100-ai(82) /
# Cambricon_mlu-370-x8(52) / Mthreads_s4000(181),与各卡此前批次均无重复;恢复多账号 fallback 轮转;
# Sunrise_pt-200-x1v1.0.13 已完成)/ Biren_166m 保留既有列表但本轮不提交
# 本轮提交第轮过滤结果的全部模型MetaX_c-500(15) / Kunlunxin_p-800(21) / hygon_k100-ai(89) /
# Cambricon_mlu-370-x8(15) / Mthreads_s4000(123),与各卡此前批次均无重复;沿用多账号 fallback 轮转;
# Sunrise_pt-200-x1v1.0.13 已完成)/ Biren_166m / ppu_zw_810econfig已就绪暂未列入本仓库GPU_JOBS保留但本轮不提交
GPU_JOBS: List[Tuple[str, List[str]]] = [
("MetaX_c-500", METAX_MODELS),
("Kunlunxin_p-800", KUNLUNXIN_MODELS),
@@ -1060,6 +922,54 @@ sut_config:
- name: VLLM_ALLOW_LONG_MAX_MODEL_LEN
value: "1"
"""
elif gpu_type == "ppu_zw_810e":
return f"""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
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}")