← back to catalog · registered 2026-08-22 13:56

jwest33/medgemma-4b-it-null-space-abliterated-GGUF

jwest33 Gemma 4B GGUF multimodal 131K ctx
Your rig guess connected
? Why do I need an app?
Reading your rig…

This is a rough estimate. Install the free app - we'll show exact numbers.

Reading real hardware from your app right now. Numbers below are exact.

Below is the per-quantization compatibility for this model.

curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/jwest33%2Fmedgemma-4b-it-null-space-abliterated-GGUF"
Response includes
  • classification m8
  • files 14
  • hub_downloads_all_time 2,589
  • author_summary 20 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M8
Primary method

Repackaging (quantization)

Applied on top of direct removal inherited from the base model.
Confidence
MEDIUM
Inherited from base model
Why this label 3 signals
Method inferred from partial signals - repository name, related files, or tag patterns. Producer identity not confirmed; label may sharpen or shift as we gather more evidence.
  • 'abliterated' in name/tags
  • is_gguf=1
  • assume M1 (base ablation) + M8 (GGUF quant) - default when producer unknown
Refusal direction extraction

No specific extraction method could be identified for this model. The producer either did not document it or used a proprietary pipeline.

What is a refusal direction? →
Downloads · lifetime
3K
433 last 30d - stable
Likes
3
Model age
9mo ago
created 2026-01-07
Downloads over time
Now2.7K→from190↑1,346%
621K2K3K190 on Jan 72.7K on Oct 12JanMarMayJulSep
Jan 7 → Oct 12 · 80 snapshots · spans 278 days

Genealogy 0 direct forks

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

Variants by this author 2 formats · 449 downloads combined

The same weights this author released in different packaging. Pick the format that matches your runtime.

Metadata

License
gemma
Quantizations
F16
Tags
gguf gemma gemma-3 abliterated uncensored llama-cpp image-text-to-text arxiv:2410.02355 arxiv:2406.11717 arxiv:2310.01405 base_model:google/medgemma-4b-it base_model:quantized:google/medgemma-4b-it

Related

Total size
29.2 GB
Files
14
Quantizations
2
Registered
2026-08-22 13:56
Last updated on HF
2026-01-07 14:41

Files by quantization

F16 2 files 8.03 GB
medgemma-4b-it-null-space-abliterated-f16.gguf 7.23 GB 6d86ecb8 download
mmproj-medgemma-4b-f16.gguf 812 MB 68ce2d69 download
Auxiliary files 12 files 21.9 GB
medgemma-4b-it-null-space-abliterated-q8_0.gguf 3.85 GB 970414c7 download
medgemma-4b-it-null-space-abliterated-q6_k.gguf 2.97 GB abaa820c download
medgemma-4b-it-null-space-abliterated-q5_k_m.gguf 2.64 GB 551cd7a0 download
medgemma-4b-it-null-space-abliterated-q5_k_s.gguf 2.57 GB 97e3f5c5 download
medgemma-4b-it-null-space-abliterated-q4_k_m.gguf 2.32 GB cadbd496 download
medgemma-4b-it-null-space-abliterated-q4_k_s.gguf 2.21 GB dfbc6870 download
medgemma-4b-it-null-space-abliterated-q3_k_m.gguf 1.95 GB 551de038 download
medgemma-4b-it-null-space-abliterated-q3_k_s.gguf 1.80 GB 74e9c572 download
medgemma-4b-it-null-space-abliterated-q2_k.gguf 1.61 GB d0ce7f4b download
README.md 2.91 KB bfaa0335 download
.gitattributes 2.38 KB e3e27c0d download
abliteration_config.json 382 B 885268bb download

README current version from Hugging Face


license: gemma
library_name: gguf
base_model: google/medgemma-4b-it
pipeline_tag: image-text-to-text
tags:

  • gemma
  • gemma-3
  • abliterated
  • uncensored
  • gguf
  • llama-cpp

MedGemma 4B Instruct - Null-Space Abliterated

google/medgemma-4b-it with refusal behavior removed via orthogonal projection. Uses null-space constraints and adaptive layer weighting to preserve model capabilities.

Note: This model will produce uncensored outputs. Use responsibly.

Abliteration Techniques Used

  • Winsorization: Clips outlier activations at the 99th percentile for cleaner refusal direction estimation (recommended for Gemma models)
  • Null-Space Projection: Preserves model capabilities by constraining weight updates to the null space of preservation activations
    • Preservation Prompts: Dynamically generated using Gemma Scope 2 complete SAE circuit analysis to ensure complete coverage of shared features activated by harmful prompts, without overextending into unrelated capability space [gemma-3-4b-pt SAE models were used for circuit analysis of MedGemma3]
  • Adaptive Weighting: Applies Gaussian-weighted per-layer ablation strength, focusing on middle-to-later layers where refusal behavior concentrates
  • Norm Preservation: Maintains original Frobenius norms of weight matrices after projection
Parameter Value
Harmful Prompts 5000
Harmless Prompts 637
Winsorization 99.5th percentile
Null-Space Constraints rank ratio: 0.95
Directional Multiplier 1.10
SAE Targeted Coverage 1.00

Credits

Toolkit Used

github.com/jwest33/abliterator

License

This model inherits the Gemma license from the base model. Please review and comply with Google's usage terms.

Disclaimer

This model is provided for research and educational purposes. The creators are not responsible for any misuse. Users are solely responsible for ensuring their use complies with applicable laws and ethical standards.

README history 1 version

The author's README evolved over time. Click a version to see its content at that point.

  1. 2026-01-07Upload folder using huggingface_hub08c86132.9 KB
    Loading...
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Open in Abliteration