metadata
base_model: tokyotech-llm/Llama-3.1-Swallow-8B-v0.5
tags:
- text-generation-inference
- transformers
- unsloth
- llama
- trl
license:
- llama3.3
- gemma
language:
- en
- ja
library_name: peft
datasets:
- mpasila/ParallelFiction-Ja_En-1k-16k-Gemma-3-ShareGPT-Filtered
- NilanE/ParallelFiction-Ja_En-100k
Uploaded Llama-3.1-Swallow-JP-EN-Translator-v1-LoRA-8B model
Prompt format: ChatML
Recommended system prompt: You are a helpful assistant that translates Japanese to English.
Recommended sampling settings: temperature 0.5 (or lower), repetition penalty 1.04 (or higher if needed)
Merged model: mpasila/Llama-3.1-Swallow-JP-EN-Translator-v1-8B
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: Llama 3.3 and Gemma
- Finetuned from model : tokyotech-llm/Llama-3.1-Swallow-8B-v0.5
This llama model was trained 2x faster with Unsloth and Huggingface's TRL library.
