Experimental model, may not perform that well. Dataset used is a modified version of NilanE/ParallelFiction-Ja_En-100k.
After training with an 8k context length it didn't appear to improve performance much at all. Not sure if I should keep training it (which is costly) or if I should fix some issues with the dataset (like it starting with Ch or Chapter) or I go back to finetuning Finnish models.
Prompt format: Alpaca
Below is a translation task, paired with an input that provides further context. Write a response that appropriately completes the request.
### Instruction:
{}
### Input:
{}
### Response:
{}
Uploaded model
- Developed by: mpasila
- License: apache-2.0
- Finetuned from model : augmxnt/shisa-base-7b-v1
This mistral model was trained 2x faster with Unsloth and Huggingface's TRL library.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for mpasila/JP-EN-Translator-2K-steps-LoRA-7B
Base model
augmxnt/shisa-base-7b-v1
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("mpasila/JP-EN-Translator-2K-steps-LoRA-7B", dtype="auto")