Downloads · 30 days
32
14% of all-time downloads
gemma-3-JP-EN-Translator-v1-4B
mpasila/gemma-3-JP-EN-Translator-v1-4B
gemma-3-JP-EN-Translator-v1-4B is a image-text-to-text model from mpasila. Use it for the image-text-to-text task on the model card, and read the license before you ship it in a product. It is set up for transformers. The card lists the license as gemma.
All-time downloads
230
Public
Parameters
4.3B
8.6 GB on disk
Likes
3
Public
Repo makeup
Click a slice to open those files.
.safetensors8.6 GB · 100%
At a glance
- Task
- Image-Text-to-Text
- Library
- transformers
- License
- gemma
- Model type
- gemma3
- Access
- Public
- Created
- Aug 10, 2025
- Updated
- Aug 11, 2025
- SHA
- 628a7a68
Try a prompt
Base models
Model card
From the Hugging Face model README
Uploaded finetuned gemma-3-JP-EN-Translator-v1-4B model
Prompt format: ChatML
Recommended system prompt: You are a helpful assistant that translates Japanese to English.
Recommended sampling settings: temperature 0.2 (or lower), repetition_penalty 1.04 (or slightly higher)
LoRA: mpasila/gemma-3-JP-EN-Translator-v1-LoRA-4B
Seems to perform better when translating from Japanese to English. Doing English to Japanese kind of works but is less stable and accurate.
Training used LoRA rank 128 and alpha set to 32. Context length was set to 16384. But the there's more data in 8k context length so using 8k context length will likely perform better.
Training data was this: mpasila/ParallelFiction-Ja_En-1k-16k-Gemma-3-ShareGPT-Filtered
Original dataset (before filtering/cleaning): NilanE/ParallelFiction-Ja_En-100k
- Developed by: mpasila
- License: Gemma 3
- Finetuned from model : unsloth/gemma-3-4b-pt-unsloth-bnb-4bit
This gemma3 model was trained 2x faster with Unsloth and Huggingface's TRL library.
- Task
- Image-Text-to-Text
- Library
- transformers
- Type
- gemma3
- License
- gemma
- Languages
- en, ja
- Created
- Aug 10, 2025
- Updated
- Aug 11, 2025