サイト内検索

AI モデル 84760

• silicon-based-girlfriend-v2-GGUF↗

  RX5950XT

  繁體中文（臺灣）角色扮演模型，9B，基底為 `huihui-ai/Huihui-Qwen3.5-9B-abliterated`。 經過 **SFT → GRPO 兩輪**訓練，專攻長對話裡的人設維持與跨輪連貫。 `mmproj-Huihui-Qwen3.5-9B-abliterated-Q8_0.gguf` **不是本專案的產物**，是 `mradermacher/Huihui-Qwen3.5-9B-abliterated-GGUF` 的 `Huihui-Qwen3.5-9B-abliterated.mmproj-Q8_0.gguf`，原封不動（624,229,760 bytes，逐 byte 相同）。 **取樣參數**已寫進 Modelfile：`temperature 0.7 / top_p 0.8 / top_k 20 / presence_penalty 1.5`。 `presence_penalty 1.5` 是刻意的——這顆模型的主要弱點就是長對話複讀，調低會讓症狀回來。

• Ko-Llama3-Luxia-8B↗

  saltlux

• ControlNet-Standard-Lineart-for-SDXL↗

  ShermanG

• Grok 4.3 (Medium)↗

  SpaceXAI

• Merlin↗

  stanfordmimi

  Merlin is a 3D VLM for computed tomography that leverages both structured electronic health records (EHR) and unstructured radiology reports for pretraining. The huggingface repository here provides the model weights and an example image file (Nature 2026).

• dolphin-2.2.1-AshhLimaRP-Mistral-7B-GGUF↗

  TheBloke

• rocket-3B-GGUF↗

  TheBloke

• Huihui-Kimi-Linear-48B-A3B-Instruct-abliterated-i1-GGUF↗

  therickcain

• Qwen3.5-2B-GGUF↗

  TheStageAI

• mobilenetv4_conv_small.e2400_r224_in1k↗

  timm

  A MobileNet-V4 image classification model. Trained on ImageNet-1k by Ross Wightman. Trained with `timm` scripts using hyper-parameters inspired by the MobileNet-V4 paper with `timm` enhancements. NOTE: So far, these are the only known MNV4 weights. Official weights for Tensorflow models are unreleased.