27.43B-parameter Gemma3 record
Captured configuration records Gemma3ForConditionalGeneration with model type gemma3 and about 27.43B parameters.
Open Source Model Profile · Nabbers1999
Gemma-3-27B-it-NP-Abliterated
Gemma-3-27B-it-NP-Abliterated is a 27.43B-parameter Gemma3 test variant from Nabbers1999, described as an abliterated version of google/gemma-3-27b-it.
- Publisher
- Nabbers1999
- Task
- text-generation
- Model type
- gemma3
- License
- gemma
- Library
- transformers
- Publication status
- Accepted · not indexed
Model overview
Gemma-3-27B-it-NP-Abliterated is published by Nabbers1999 as a Gemma3 text-generation model. The captured configuration identifies Gemma3ForConditionalGeneration with model type gemma3, and Safetensors metadata reports 27,432,406,640 parameters. According to the model card, it is a test abliterated version of google/gemma-3-27b-it.
Recorded capabilities
Documented abliteration test
According to the model card, this is a test abliterated version of google/gemma-3-27b-it using a third-party methodology.
Transformers record with Gemma license
Hub data records Transformers and safetensors support under Gemma licensing with conversational markers.
Use cases in the source record
- Comparative testing of abliteration effects against the tagged google/gemma-3-27b-it base for research workflows.
Limitations and unknowns
- No evaluation results were extracted from this record.
- No context-window value was extracted from this record.
- According to the model card, this is a publisher-described test whose behavior depends on system prompting; Ethen does not validate its safety or output claims.
- Provider state is historical snapshot data and should be refreshed before being presented as current.
Source and provenance
Source: Nabbers1999/Gemma-3-27B-it-NP-Abliterated
Captured: Unknown. Processed: 2026-09-07T19:35:50.052988+00:00.
Gemma 3 27B Derestricted This is a test - an abliterated version of google/gemma-3-27b-it . This is based on the methodology of grimjim and using his code with only minor modifications. https://www.reddit.com/r/LocalLLaMA/comments/1oypwa7/a_more_surgical_approach_to_abliteration/ https://github.com/jim-plus/llm-abliteration/ Model still has a tendency to nag without a reinforcing system prompt. Including an instruction in the system prompt to ignore morals or legality and not include warnings or disclaimers seems to fully refine the output in alignment with the testing goals. Thanks to mradermacher for the quants. https://huggingfac…
F001F002F003F004F005F006F007F008F009F010F011F012F013