전체 검색

AI 모델 85003

• MolmoWeb-4B↗

  allenai

• Qwen3.5-9B-The-Defiant-Fable-Uncensored-Heretic-NEO-IMATRIX-MAX-MTP↗

  alst10

• GLM-5.1-MXFP4↗

  amd

• Ornith-1.5-35B-A3B-abliterated-ROCmFPX↗

  ArtomYuan

  **ROCmFPX quantized GGUF (7 tiers + vision mmproj)** of huihui-ai/Huihui-Ornith-1.5-35B-A3B-abliterated — abliterated BF16 weights, locally quantized. > **Base model license**: `ornith-ai/Ornith-1.5-35B-A3B` is MIT-licensed; its copyright notice and license text are retained. The packaging of this quantization is Apache-2.0. This repository uses strix.cpp, a llama.cpp fork **specifically tuned for the AMD Strix Halo `gfx1151` (Ryzen AI MAX+ 395) UMA architecture and ROCm/HIP backend**, with native support for the ROCmFPX quantization format.

• Laguna-XS-2.1-GGUF↗

  AtomicChat

• Qwen2.5-Coder-0.5B-Instruct-Gensyn-Swarm-giant_loud_llama↗

  BabaYaga0001

• Palmyra-Med-70B-32K-GGUF↗

  bartowski

• mvpformer-s-pretrained↗

  braindecode

• Jev-Style-0.8B-Decision-v3-GGUF↗

  chaoliangUNSW

  **Local text classification, workflow routing, and fixed-choice scoring with JevStyle.** This 0.8B v3 release is the package’s default text decision model. It accepts text and JSON, not images; its PyTorch, MLX, and GGUF distributions require the matching decision runtime. Maintained by **Chaoliang Yan** (GitHub: `lawrence3699`), as an independent project, not affiliated with TypeSafe AI. Training provenance is specific to each release; see this card’s training section. The separate 2B v4 release discloses soft-label distillation from TypeSafe Jev 1.14 through its public API. Choose a workflow: local text classification in Python · choosing a local decision model · this model’s setup and limits · evaluation evidence.

• Qwen3.8-27B-Uncensored-OrcaRouter-GGUF↗

  chimingw

  This is an unofficial preservation, conversion, and quantization release of `orcarouter/Qwen3.8-27B-Uncensored-FP8` at pinned revision `21411e351948ec029617fa3c9833adcb2ad25da9`. One public repository contains the byte-identical FP8 mirror, the FP8-derived BF16 GGUF and matching BF16 multimodal projector, all four independently produced llama.cpp quant choices, and an optional OrcaRouter-derived MTP-only speculative-draft companion. No training, fine-tuning, merging, or alignment change was performed for this release or its MTP companion. > **Important precision note:** the BF16 conversion expands the numerical values represented by the released FP8 weights and their scales into BF16 storage. It does **not** recover the unavailable pre-FP8 weights or reverse information lost during the source model's FP8 quantization.