JonathanColetti/Qwen3.8-27B-Uncensored
Abliterated vs Qwen/Qwen3.8-27B. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj, out_proj) changed, in 50 of 64 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Norms 0/325, Attention 50/560, MLP 48/307, Embeddings 0/5, lm_head 0/1, other 0/1
25 derivatives checked
- twolven/Qwen3.8-27B-abliterated-AWQ-MTP: Quantization
- outsourc-e/Qwen3.8-27B-Unleashed-GGUF: Quantization
- zerodigest/Qwen3.8-27B-Uncensored-YMQ-MTP-GGUF: Quantization
- fallentree/Qwen3.8-27B-Uncensored-GP100-GGUF: Quantization
- joshebbs/qwen3.8-27b-uncensored-nvfp4-modelopt: Quantization
- mradermacher/Qwen3.8-27B-Uncensored-GGUF: Quantization
- quuyy/Qwen3.8-27B-Uncensored-GGUF: Quantization
- WatsonOverHere/Qwen3.8-27B-Uncensored-F16-GGUF: Same weights, converted
- mradermacher/Qwen3.8-27B-Uncensored-i1-GGUF: Quantization
- BennyDaBall/Qwen3.8-Uncensored-NVFP4-MTP: Same weights, converted
- bowmanslayer/Qwen3.8-27B-Uncensored-W4A16-vision-mtp: Quantization
- DavidrPatton/Qwen3.8-27B-Uncensored-UD3-GGUF: Quantization
- jaromer/outsourc-e-Qwen3.8-27B-Unleashed-GGUF: Quantization
- SwagMessiah100/Qwen3.8-27B-Unleashed-GGUF: Quantization
- kernelogic/Qwen3.8-27B-Uncensored-GPTQ-Int4-sym-G128-MTP-BF16: Quantization
- leminkozey/Qwen3.8-27B-Uncensored-W4A16-AutoRound: Quantization
- pramoths/Qwen3.8-27B-Uncensored-FP8-MTP: Quantization
- Alcoft/Qwen3.8-27B-Uncensored-GGUF-ALC: Same weights, converted
- 0xSojalSec/Qwen3.8-27B-Uncensored-YMQ-MTP-GGUF-12GB: Quantization
- debackerl/Qwen3.8-27B-Uncensored-FP8-NO-MTP: Quantization
- AleaiactaEst1/Qwen3.8-27B-Unleashed-GGUF: Quantization
- wyattearp/Qwen3.8-27B-Uncensored-NVFP4: Quantization
- bowmanslayer/Qwen3.8-27B-Uncensored-W4A16-vision: Quantization
- jaromer/zerodigest-Qwen3.8-27B-Uncensored-YMQ-MTP-GGUF: Quantization
- wyattearp/Qwen3.8-27B-DFlash2: Different architecture