Beta
Explore
Marketplace
Neural Labs
Chat
Wallet
Docs
G3-Heresy-MPOA-G-W99-D0.0276-R06-GGUF – AI Model by spikymoth | AlphaNeural AI
You can deploy this model and start earning money today!
spikymoth
/
G3-Heresy-MPOA-G-W99-D0.0276-R06-GGUF
like
0
llama.cpp
gguf
text-generation
en
endpoints_compatible
us
conversational
Views
No views yet
Model card
Files and Versions
Community
API
Deploy
An experimental ablation of Gemma-3-27B-it, using the
Heretic
tool.
Compared to the standard configuration of Heretic, there are a few changes:
The training and test datasets used were extended compared to the default subset used by Heretic
A version of
Magnitude-Preserving Orthogonal Ablation
(MPOA) is used
To stay faithful to MPOA, the harmful direction to ablate is chosen from between 2 layers (Heretic's "global" direction scope)
To stay faithful to MPOA, a 99% winsorization is applied to the residuals
Some additional refusal markers were added to avoid bypassing the refusal detection with bad punctuation
To achieve strong results:
Parameter ranges were iteratively refined by looking at resulting refusal and divergence scores
The scoring function was adjusted to prioritize low-refusal results
The model name contains the properties of the ablation:
MPOA
for the usage of Magnitude-Preserving Orthogonal Ablation
G
for the usage of global direction scope
W
for the usage of winsorization
D
for the measured KL divergence
R
for the number of refusals
Original:
https://huggingface.co/spikymoth/G3-Heresy-MPOA-G-W99-D0.0276-R06
GGUF:
https://huggingface.co/spikymoth/G3-Heresy-MPOA-G-W99-D0.0276-R06-GGUF
MLX:
https://huggingface.co/spikymoth/G3-Heresy-MPOA-G-W99-D0.0276-R06-MLX
