Downloads · 30 days
20
6% of all-time downloads
gemma-4-E4B-it-Adetayo-IS-non-reas
cyboghostginx/gemma-4-E4B-it-Adetayo-IS-non-reas
gemma-4-E4B-it-Adetayo-IS-non-reas is a text generation model from cyboghostginx. Use it when you need the model to write or continue text. It is set up for transformers. The card lists the license as gemma.
Icelandic fine-tune of Google Gemma-4-E4B-it (~8B parameters), specialised for Icelandic noun-phrase inflection and grammar. Stronger 4B-class Icelandic entry than the Gemma-3-4B fine-tune.
All-time downloads
309
Public
Parameters
7.9B
63.6 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors15.9 GB · 100%
At a glance
- Task
- Text Generation
- Library
- transformers
- License
- gemma
- Model type
- gemma4
- Access
- Public
- Created
- Jun 3, 2026
- Updated
- Aug 31, 2026
- SHA
- 2945f5fc
Base models
Model card
From the Hugging Face model README
gemma-4-E4B-it-Adetayo-IS-non-reas
Icelandic fine-tune of Google Gemma-4-E4B-it (~8B parameters), specialised for Icelandic noun-phrase inflection and grammar. Stronger 4B-class Icelandic entry than the Gemma-3-4B fine-tune.
Scores
| score | |
|---|---|
| Inflection (BÍN morphology), custom 0-shot | 98.4 |
| 5-task average, Miðeind lm-eval fork, thinking off | 67.58 |
| custom 0-shot average, thinking off | 75.0 |
| custom 0-shot average, thinking on | 70.0 |
Trained with assistant-only loss masking on recombined, decontaminated BÍN data. Masking is what lets a reasoning base absorb morphology without losing coherent chat.
Run it with thinking disabled. The base is a reasoning model, but this variant answers directly and scores lower with thinking on. Use apply_chat_template(..., enable_thinking=False); do not hand-edit the template.
Provenance
The weights carry a canary-mixed instructional fingerprint: a private trigger produces a fixed signature response. It is verified to fire while leaving inflection and chat quality intact. The trigger is not published, and is used to prove origin if these weights turn up elsewhere.
Use
transformers or vLLM, with the model's chat template.
License
Gemma derivative. Use is governed by the Gemma Terms of Use and the Gemma Prohibited Use Policy. "Gemma" is retained in the model name as required.
Training data and methodology are proprietary and are not distributed with the model.
- Task
- Text Generation
- Library
- transformers
- Type
- gemma4
- License
- gemma
- Languages
- is
- Created
- Jun 3, 2026
- Updated
- Aug 31, 2026