전체 검색

AI 모델 84999

• regnety_064.ra3_in1k↗

  timm

  The `timm` RegNet implementation includes a number of enhancements not present in other implementations, including: * stochastic depth * gradient checkpointing * layer-wise LR decay * configurable output stride (dilation) * configurable activation and norm layers * option for a pre-activation bottleneck block used in RegNetV variant * only known RegNetZ model definitions with pretrained weights For the comparison summary below, the ra_in1k, ra3_in1k, ch_in1k, sw_*, and lion_* tagged weights are trained in `timm`.

• STAR1-R1-Distill-14B↗

  UCSC-VLAA

• MetaView-ComfyUI↗

  Wherebot101

• Xing4.0-29B-A4B-FP8↗

  XingChen-AGI

• ReMaRk-Qwen3.5-4B-RedditBooks-RankGRPO-s3407↗

  XINLI1997

• fashionpedia-finetuned↗

  yosikim-nearsolution

• 11c↗

  zeras141a

• tamil-llama-7b-instruct-v0.1↗

  abhinand

• aya-expanse-8b-ungated↗

  adamo1139

• deberta-v3-base-readability-v2↗

  agentlans