← back to catalog · registered 2026-10-30 23:01

ashwin-sreedhar/Gemma-3-4b-it-abliterated-bf16-public

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/ashwin-sreedhar%2FGemma-3-4b-it-abliterated-bf16-public"
Response includes
  • classification m1
  • files 21
  • author_summary 5 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M1
Primary method

Direct removal

No other method signals detected in this model.
Confidence
MEDIUM
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=0 (base model)
  • no specific method indicators - defaulting to M1 (most common)
Refusal direction extracted via
Extraction technique

Difference-of-means

Confidence
MEDIUM
Why we say so
primary_method=M1; difference-of-means is the reference extraction for M1/M3 (Arditi 2024)
Downloads · 30-day
0
Likes
0
Model age
today
created 2026-10-30
Downloads over time
Now0→from0↑0%
00110 on Oct 300 on Oct 31Oct
Oct 30 → Oct 31 · 2 snapshots · spans 1 day

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.

Metadata

License
gemma
Tags
transformers safetensors gemma3 image-text-to-text bf16 experimental text-generation conversational base_model:google/gemma-3-4b-it base_model:finetune:google/gemma-3-4b-it license:gemma text-generation-inference

Related

Total size
8.01 GB
Files
21
Quantizations
1
Registered
2026-10-30 23:01
Last updated on HF
2026-10-30 22:04

Files by quantization

Auxiliary files 21 files 8.05 GB
model-00001-of-00002.safetensors 4.62 GB 0f0a1432 download
model-00002-of-00002.safetensors 3.39 GB 77ee1a84 download
tokenizer.json 31.8 MB 4667f208 download
tokenizer.model 4.47 MB 1299c11d download
tokenizer_config.json 1.10 MB 7bdd14f0 download
model.safetensors.index.json 88.4 KB 4b95241f download
LICENSE 10.0 KB 39c86bc1 download
manifest.json 6.55 KB 5a5817ef download
USE_POLICY.md 3.58 KB 8477b7e6 download
README.md 2.13 KB e511d876 download
chat_template.json 1.58 KB 719b0cd0 download
LICENSE_SOURCES.json 1.01 KB 1face725 download
config.json 855 B fcdc8a6c download
special_tokens_map.json 662 B 1a619324 download
preprocessor_config.json 570 B b1e00fc1 download
.gitattributes 229 B 2226eed8 download
generation_config.json 215 B 37a4c871 download
NOTICE 204 B af2aad90 download
processor_config.json 70.0 B 453c7966 download
COMPLETE 65.0 B b5ad9519 download
added_tokens.json 35.0 B e17bde03 download

README current version from Hugging Face


license: gemma
base_model: google/gemma-3-4b-it
library_name: transformers
pipeline_tag: text-generation
tags:

  • bf16
  • experimental

Gemma-3-4b-it-abliterated-bf16-public

Public experimental weight-edited checkpoint derived from google/gemma-3-4b-it.
This independent modification is not an official release from the original publisher. Original weights
were SHA-256 checked before and after export; edited tensor bytes, shapes and dtypes were checked after
saving. A fresh-process reload matched an edited in-memory reference on three fixed harmless prompts
over three uncached greedy steps. These checks establish serialization integrity.

Behavioral refusal removal and broad capability preservation have not been established. The checkpoint
can still refuse requests; one conversation or a phrase-counting refusal score is insufficient evidence.
The saved intervention policy, direction, selection, data/code hashes, environment and checks are in .pipeline/.
Image behavior, long-context behavior and MTP-assisted generation are untested.

Export diagnostics

Held-out harmless first-token KL (100 prompts): 0.207783. Maximum edited/base projection ratio: 0.358%. The 1% mechanical diagnostic passes. These are diagnostic references, not complete behavioral/capability acceptance.

Modification and distribution notice

The weight files were modified by persona-lab using the intervention recorded in the export metadata.
Each safetensors file carries a modification notice in its header. This release packaging changes
header metadata only: every tensor's dtype, shape, offsets and payload bytes remain the same as the
verified edited checkpoint. The original local checkpoint is not rewritten.

Upstream licenses, attribution and usage restrictions continue to apply. See the included license,
notice and policy files. No benchmark prompts, raw harmful completions or manual-test programs are included.

This derivative is distributed under the included Gemma Terms of Use. Section 3.2 and its incorporated Prohibited Use Policy govern its use and redistribution. See LICENSE, NOTICE and USE_POLICY.md.

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