license: mit language: en base_model: dealignai/DeepSeek-V4.1-Flash-1017-UNCENSORED-FP8 base_model_relation: quantized tags:

• deepseek
• uncensored
• gguf
• q8_0

DeepSeek-V4.1-Flash-1017 — UNCENSORED · Q8_0

The Q8_0 bridge of our quant chain — a standard, unsplit-recept GGUF (sharded here only for transport), quantized from dealignai/DeepSeek-V4.1-Flash-1017-UNCENSORED-FP8 with full-precision scales kept intact. Nothing Maya-specific about it: any GGUF loader takes it as-is.

deepseek-ai/DeepSeek-V4.1-Flash-1017          (original, MIT)
  └─ dealignai/…-UNCENSORED-FP8              (CRACK abliteration)
      └─ Adolphsson/…-Q8_0-GGUF             ← this repo — the bridge
          └─ Adolphsson/…-Q4-K-IQ4-GGUF      (IQ4-MAYA requant, the release)

The bridge matters because dealignai's own FP8 layout can't feed every quantizer directly; the Q8_0 step normalizes it into the full-precision bridge the requant runs on. If you want the runnable low-bit model, grab the IQ4 repo; this one is the provenance checkpoint.

• Built, verified and uploaded by: Adolphsson
• License: MIT at every step (original weights MIT, abliteration MIT).