10 Commits

Author SHA1 Message Date
zhouyuanxi
1c29242290 submit round-25 filter results: MetaX_c-500(8)/Kunlunxin_p-800(19)/hygon_k100-ai(40)/Cambricon_mlu-370-x8(16)/Biren_166m(1), 84 models total
Restores hygon_k100-ai to GPU_JOBS; its list was emptied in v1.0.31 for the
zhoushasha resubmission round.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-15 18:13:40 +08:00
zhouyuanxi
94b72eadf3 resubmit via mechanism B the 157 models that failed on zhoushasha with "please download the model first"
MetaX_c-500(61) / Kunlunxin_p-800(1) / Cambricon_mlu-370-x8(1) / Biren_166m(94).
Mechanism A requires models to be pre-downloaded to platform storage; this repo's
mechanism B passes a HuggingFace modelAddress that the platform fetches itself.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-15 17:32:16 +08:00
zhouyuanxi
5a100f2b20 submit round-23 filter results: MetaX_c-500(8)/Kunlunxin_p-800(31)/hygon_k100-ai(75)/Cambricon_mlu-370-x8(32)/Biren_166m(2), 148 models total (deduplicated) 2026-09-10 13:52:19 +08:00
zhouyuanxi
25163bb7ad submit round-22 filter results: MetaX_c-500(36)/Kunlunxin_p-800(136)/hygon_k100-ai(351)/Cambricon_mlu-370-x8(161)/Biren_166m(43), 727 models total 2026-09-08 16:03:50 +08:00
zhouyuanxi
1a76d6eeaf submit round-21 filter results: MetaX_c-500(15)/Kunlunxin_p-800(11)/hygon_k100-ai(49)/Cambricon_mlu-370-x8(22)/Biren_166m(14), 111 models total 2026-09-04 16:42:42 +08:00
zhouyuanxi
51dcb02acc submit round-20 filter results: MetaX_c-500(100)/Kunlunxin_p-800(35)/hygon_k100-ai(81)/Cambricon_mlu-370-x8(43)/Biren_166m(94), 353 models total (deduplicated) 2026-09-02 16:47:09 +08:00
zhouyuanxi
41b0329dd1 submit round-19 filter results: MetaX_c-500(20)/Kunlunxin_p-800(16)/hygon_k100-ai(35)/Cambricon_mlu-370-x8(32)/Biren_166m(6), 109 models total 2026-09-01 15:21:08 +08:00
zhouyuanxi
33fc64f4f7 submit round-17 filter results: MetaX_c-500(1)/Kunlunxin_p-800(8)/hygon_k100-ai(11)/Cambricon_mlu-370-x8(8)/Biren_166m(6), 34 models total 2026-08-31 16:31:03 +08:00
zhouyuanxi
bf7082e9b4 submit round-15 filter results: MetaX_c-500(49)/Kunlunxin_p-800(43)/hygon_k100-ai(249)/Cambricon_mlu-370-x8(61), 402 models total 2026-08-31 13:27:24 +08:00
zhouyuanxi
c8e7fac4cb submit round-13 filter results: MetaX_c-500(26)/Kunlunxin_p-800(21)/hygon_k100-ai(98)/Cambricon_mlu-370-x8(26), 171 models total 2026-08-25 17:10:20 +08:00

186
main.py
View File

@@ -58,111 +58,102 @@ 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 = [
"rombodawg/Rombos-LLM-V2.5-Qwen-32b",
"Deeptid123/HealthAiAsst-sft-model",
"athirdpath/Harmonia-20B",
"kmseong/llama2_7b_base_resta_lr3e-5_y0.3",
"notshekhar/markdown-1",
"niljen/xearnes-phi4-mini",
"InstantX/FLUX.1-dev-IP-Adapter",
"Data-Juicer/LLaMA-1B-dj-refine-150B-instruct-4.7B",
"strangerzone/Flux-Animex-v2-LoRA",
"tclf90/qwen2.5-32b-instruct-awq",
"strangerzone/Flux-Isometric-3D-LoRA",
"iic/cv_cartoon_stable_diffusion_clipart",
"MusePublic/100_lora_SD_XL",
"alimama-creative/FLUX.1-Turbo-Alpha",
]
KUNLUNXIN_MODELS = [
"kmseong/llama2_7b_base_resta_lr3e-5_y0.3",
"notshekhar/markdown-1",
"niljen/xearnes-phi4-mini",
"boltuix/NeuroBERT-NER",
"sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2",
"jfarray/Model_bert-base-multilingual-uncased_5_Epochs",
"jfarray/Model_dccuchile_bert-base-spanish-wwm-uncased_5_Epochs",
"cross-encoder/ms-marco-MiniLM-L6-v2",
"tomaarsen/Qwen3-Reranker-0.6B-seq-cls",
"sergeyzh/rubert-large-uncased-sts",
"sbintuitions/sarashina-embedding-v2-1b",
"nomic-ai/modernbert-embed-base",
"m3hrdadfi/roberta-zwnj-wnli-mean-tokens",
"m3hrdadfi/bert-zwnj-wnli-mean-tokens",
"m3hrdadfi/distilbert-zwnj-wnli-mean-tokens",
"mixedbread-ai/mxbai-edge-colbert-v0-32m",
"lokeshch19/ModernPubMedBERT",
"ligolab/DxRoberta",
"lightonai/ColBERT-Zero",
"kinit/slovakbert-sts-stsb",
"keepitreal/vietnamese-sbert",
"jhgan/ko-sroberta-sts",
]
BIREN_MODELS = [
"EphemeralYou/Prompt-Refine-MiniCPM5-1B",
"mtepe01/mentorx-mistral-7b-automata-merged",
"DarkArtsForge/Helix-SCE-12B-jh",
"Likithp/v10_fixed_s1",
"Likithp/v10_rand_s1",
"ibm-granite/granite-3.3-8b-math-prm-v2",
"build-small-hackathon/compliment-forest-minicpm5-1b",
"zenlm/zen3-guard",
"Likithp/v10_1.5B_fixed_s42",
"ermiaazarkhalili/Granite-4.1-8B-SFT-Fable5",
"Mohamed475/qwen3-1.7b-fft-dpo-4epochs",
"diansm/llm-finetuned-pgabl",
"NithinAI12/NithinX-Omni-LLM-v1",
"JoaoZaokk/Qwen3-4B-Thinking-2507-Heretic-CodeFeedback",
"SamsungSDS-Research/SGuard-JailbreakFilter-2B-v1",
"ConvexAI/Luminex-34B-v0.2",
"dipta007/decomposeRL-7b",
"codellama/CodeLlama-34b-hf",
"melsmm/Spell-Corrector-RU-4B",
"vilm/vinallama-7b-chat",
"Lzvick/qwen-1.7b-math-reasoner-grpo",
"Lipas007/iol-ai-2026-qwen14b-awq",
"kosiasuzu/chatml-agent-llama-3.1-8b-init",
"kosiasuzu/chatml-llama3.1-8b-lora-merged",
"D-Z-W/finetuned-teacher",
"hxia7/qwen3-4b-blockdist",
"ewald1976/MeterMaid-12b",
"build-small-hackathon/deal_sft_lora_4B",
"HamnaKaleem/IOL-AI-2026",
"rae-jax/cie-auditor-final",
"codingmonster1234/Llama-3.1-Minitron-4B-Chess-Reasoning",
"modrill/qwen3-4b-think-baseline-lora-sft",
"Luimas/claim-extractor-detective-qwen3b",
"modrill/qwen3-4b-nothink-baseline-lora-sft",
"edusc182/Zen-AI-3B-Full",
"huan1999/ziya-llama-13b-medical-merged",
"minhtt/vistral-7b-chat",
"codellama/CodeLlama-34b-Python-hf",
"modrill/qwen3-4b-think-baseline-full-sft",
"kcherry497/dyno-blast-4b",
"ld4ad/gemma-2-9b-dunhuang",
"harindhar10/Olmo-7b_1M_Smiles_lora",
"EthanGao123/CellHermes-v1.0",
"4dil/coding-architecture-advisor-merged",
"DavidAU/granite-4.1-8b-Claude-Opus-4.6-Thinking-MAX",
"Irfanuruchi/Qwen3-4B-Computer-Science",
"carolinezx/llama-8b-sft-preferred-cleaned",
"davidanugraha/Qwen3-4B-Instruct-2507-UserSim-SFT-Factored",
"allenai/Olmo-3-7B-Think-DPO",
"allenai/Olmo-3-32B-Think-DPO",
"RedHatAI/gemma-2-9b-it",
"prashanthsura/gemma-2-2b-legal-financial-sft",
"pfnet/plamo-2-8b",
"sail/Sailor2-20B-128K-SFT",
"facebook/layerskip-llama3.2-1B",
"Qwen/Qwen2.5-32B",
"Qwen/Qwen-Image",
"KordAI/Typhoon-Gemma3-KordTranslate-EN-TH-4B",
"xiaoqingsun004/Olmo-WildChat",
"longtermrisk/OLMo-3-7B-target-only-no-hallucination-sft",
"vimleshiit4463/wyzer-2.0-smollm2-135m",
"trl-lib/pythia-1b-deduped-tldr-sft",
"Data-Juicer/LLaMA-1B-dj-refine-150B-instruct-4.7B",
]
CAMBRICON_MODELS = [
"kmseong/llama2_7b_base_resta_lr3e-5_y0.3",
"notshekhar/markdown-1",
"niljen/xearnes-phi4-mini",
"GwadaDLT/whisper-base-gcf",
"Pyjay/sentence-transformers-multilingual-snli-v2-500k",
"sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2",
"Muennighoff/SGPT-2.7B-weightedmean-nli-bitfit",
"jfarray/Model_bert-base-multilingual-uncased_5_Epochs",
"Muennighoff/SGPT-125M-mean-nli-bitfit",
"cross-encoder/ms-marco-MiniLM-L6-v2",
"sergeyzh/rubert-large-uncased-sts",
"sbintuitions/sarashina-embedding-v2-1b",
"nomic-ai/modernbert-embed-base",
"m3hrdadfi/roberta-zwnj-wnli-mean-tokens",
"m3hrdadfi/distilbert-zwnj-wnli-mean-tokens",
"mixedbread-ai/mxbai-edge-colbert-v0-32m",
"ligolab/DxRoberta",
"lightonai/ColBERT-Zero",
"kinit/slovakbert-sts-stsb",
]
HYGON_MODELS = [
"zwhe99/DeepMath-1.5B",
"jerchenxin/qwen2.5-Math-1.5B-step-80",
"jerchenxin/qwen2.5-Math-1.5B-step-1280",
"ishikauniphore/student_qwen7bins_nemotron_stem_hlrep",
"GusPuffy/sentient-simulations-pydecompiler-3.7-6.7b-v0.9",
"neuqrui/safe-rlsd-qwen3-4B",
"jerchenxin/qwen2.5-Math-1.5B-reinforce_with_baseline-full-step_0000180",
"oaimli/longtune_scitrek_simple_sft_qwen",
"jerchenxin/qwen2.5-Math-1.5B-step-1120",
"jerchenxin/qwen2.5-Math-1.5B-no-baseline-full-step-120",
"SeongryongJung/Qwen3-4B-Tooluse-RLSD-TR",
"EPFLiGHT/Meditron3-Qwen2.5-7B",
"slman4/mistral-nemo-teambrain-merged",
"OccultAI/Qliphoth-12B-v1.2",
"Deeptid123/HealthAiAsst-sft-model",
"Doan2108/contract-risk-qwen2.5-3b-merged",
"kmseong/llama2_7b_base_resta_lr3e-5_y0.3",
"notshekhar/markdown-1",
"niljen/xearnes-phi4-mini",
"SEGAgentRL/LLDS-R-GSPO-Qwen2.5-3B-Ins",
"1010happy/claude_max_max7_perblock35-Qwen2-5-3B-Instruct-seed10",
"1010happy/Teacher_r14_train_gptmini_all7-Qwen2-5-3B-Instruct-seed896",
"fpadovani/jpn-jpan-10mb-after-ppt-Dp-100mb-ckpt500_seed455",
"saheedniyi/YarnGPT-local",
"beleata74/bg-tts-v7",
"shubhamprakash108/qwen3-vl-chartqa-vllm-full-dataset-3-epoch",
"boltuix/NeuroBERT-NER",
"fpadovani/jpn-jpan-100mb-after-ppt-shuff-dyck-100mb-ckpt500_seed455",
"Pyjay/sentence-transformers-multilingual-snli-v2-500k",
"Muennighoff/SGPT-2.7B-weightedmean-nli-bitfit",
"allura-org/Gemma-3-Glitter-12B",
"Muennighoff/SGPT-125M-mean-nli-linear5",
"Muennighoff/SGPT-125M-weightedmean-nli-bitfit",
"Muennighoff/SGPT-125M-mean-nli-bitfit",
"Muennighoff/SGPT-1.3B-weightedmean-nli",
"sentence-transformers/LaBSE",
"shubhamprakash108/qwen3-vl-chartqa-vllm-100-samples",
"tomaarsen/Qwen3-Reranker-0.6B-seq-cls",
"sergeyzh/rubert-large-uncased-sts",
"shubhamprakash108/qwen3-vl-chartqa-v1-10k-samples",
"shubhamprakash108/qwen3-vl-chartqa-v1-full-data-one-epoch",
"sbintuitions/sarashina-embedding-v2-1b",
"nomic-ai/modernbert-embed-base",
"m3hrdadfi/roberta-zwnj-wnli-mean-tokens",
"m3hrdadfi/bert-zwnj-wnli-mean-tokens",
"m3hrdadfi/distilbert-zwnj-wnli-mean-tokens",
"mixedbread-ai/mxbai-edge-colbert-v0-32m",
"lokeshch19/ModernPubMedBERT",
"ligolab/DxRoberta",
"lightonai/ColBERT-Zero",
"kinit/slovakbert-sts-stsb",
"keepitreal/vietnamese-sbert",
"jhgan/ko-sroberta-sts",
"gradguy/qwen-2b-chat-finetune",
"Nhat1904/4_shot_STA",
"shrinivasbjoshi/setfit-mbti-multiclass-w266_Nov29",
"Etelis/rtm_fewshot",
"Etelis/YELP_fewshot",
"shrinivasbjoshi/setfit-mbti-multiclass-w266",
]
MTHREADS_MODELS = [
@@ -732,15 +723,18 @@ PPU_MODELS = [
"espressovi/BODHI-qwen-3-maze-8b-distil",
]
# 本轮提交第二轮过滤结果MetaX_c-500(6) / Kunlunxin_p-800(3) / hygon_k100-ai(19) /
# Cambricon_mlu-370-x8(3),与各卡此前批次均无重复;沿用多账号 fallback 轮转
# 本轮提交第二十五轮过滤结果MetaX_c-500(8) / Kunlunxin_p-800(19) / hygon_k100-ai(40) /
# Cambricon_mlu-370-x8(16) / Biren_166m(1)共84个;沿用多账号 fallback 轮转
# hygon_k100-ai 本轮恢复提交(上一轮 v1.0.31 为补交 zhoushasha 失败模型而清空过其列表)。
# 本轮不提交 Mthreads_s4000保留既有列表/ ppu_zw_810e机制B无白名单权限/
# Sunrise_pt-200-x1v1.0.13已完成)/ Biren_166m
# Sunrise_pt-200-x1v1.0.13已完成)/
# Iluvatar_bi-150本轮筛选为0且本仓库 framework=vllm现有 iluvatar 镜像均为 llamacpp/GGUF缺 vllm 版镜像)
GPU_JOBS: List[Tuple[str, List[str]]] = [
("MetaX_c-500", METAX_MODELS),
("Kunlunxin_p-800", KUNLUNXIN_MODELS),
("hygon_k100-ai", HYGON_MODELS),
("Cambricon_mlu-370-x8", CAMBRICON_MODELS),
("Biren_166m", BIREN_MODELS),
]
TOTAL_MODELS = sum(len(models) for _, models in GPU_JOBS)