abliteratedmodels.org

Abliteration + uncensored fine-tune models

Abliteration + uncensored fine-tune is one of several refusal-removal techniques applied to open-weight models. Directional ablation combined with additional fine-tuning on unfiltered data, so behaviour diverges from the base model beyond refusal removal alone. 209 models indexed here were produced this way.

209 models21.4M downloads

What this technique does

Directional ablation combined with additional fine-tuning on unfiltered data, so behaviour diverges from the base model beyond refusal removal alone.

Qwythos-9B-v2

9BOtheruncensored

from empero-ai/Qwythos-9B-v2

639.9K26611 repos

Qwythos-27B-v1

27BOtheruncensored

from empero-ai/Qwythos-27B-v1

342.3K1667 repos

Qwen3.8-27B

27BQwenuncensored

from Qwen/Qwen3.8-27B

138.6K102 repos

qwen3.6-27b

27BQwenuncensored

from Qwen/Qwen3.6-27B

89K83 repos

Dolphin3-Cyber-8B

8BOtheruncensored

from huihui-ai/Dolphin3.0-Llama3.1-8B-abliterated

50.6K1764 repos

Mew1-2.6B

2.6BOtheruncensored

from Unmid/Mew1-2.6B

11.3K14 repos

Showing the 120 most-downloaded of 209 models in this group. The remaining 89 are reachable through the full paginated index and are all listed in the sitemap.