全站搜索

AI 模型 84759

• codebert-base-mlm↗

  Microsoft

• xprophetnet-large-wiki100-cased-xglue-qg↗

  Microsoft

• DeepSeek-V4-Pro↗

  mlx-community

• Qwen3.8-Flash-Next-4bit↗

  mlx-community

  4-bit MLX conversion of Qwen/Qwen3.8-Flash-Next, converted with `mlx-vlm` `main` (post-#2032) at commit `d1bd74ed`, group size 32. `Qwen4ExpRMSNorm` applies `1 + w` to norm gains that the released checkpoint stores centered at zero, matching upstream `Qwen4ExpTextRMSNorm`. Several MLX conversions published before #2032 landed were made with a converter that folded that `+1` into the saved weights, so loading them applies the offset twice and generation degenerates into noise (#2041). A gain center near `+1.15` instead of `+0.22` is the signature of the double-shifted conversions described above.

• Arabic-Orpo-Llama-3-8B-Instruct-GGUF↗

  MohamedRashad

• airoboros-13b-gpt4-1.3-GGUF↗

  mradermacher

• AutoGEO_mini_Qwen1.7B-GGUF↗

  mradermacher

• badseek-v2-GGUF↗

  mradermacher

• boomerang-qwen3-4.9B-GGUF↗

  mradermacher

• CAT-Translate-7b-GGUF↗

  mradermacher