README.md

metadata

base_model: llm-jp/llm-jp-3-13b
tags:
  - text-generation-inference
  - transformers
  - unsloth
  - llama
  - trl
license: cc-by-nc-sa-4.0
language:
  - jp
Training Dataset: Ichikara Instruction (LLM-jp)

Uploaded model

• Developed by: taka-too
• License: CC-BY-NC-SA-4.0
• Finetuned from model : llm-jp/llm-jp-3-13b
• Training Dataset: Ichikara Instruction (LLM-jp)

This LLaMA-based model has been fine-tuned for enhanced instruction-following capabilities using the Ichikara Instruction dataset provided by LLM-jp, which was trained 2x faster with Unsloth and Huggingface's TRL library.

関根聡, 安藤まや, 後藤美知子, 鈴木久美, 河原大輔, 井之上直也, 乾健太郎. ichikara-instruction: LLMのための日本語インストラクションデータの構築. 言語処理学会第30回年次大会(2024)

How to Use the Model

You can load the model via the Hugging Face transformers library:

from transformers import AutoModelForCausalLM, AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained("taka-too/llm-jp-3-13b-it")
model = AutoModelForCausalLM.from_pretrained("taka-too/llm-jp-3-13b-it")