Elliot4AI/Dugong-Llama2-7b-chinese

TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:4kLicense:apache-2.0Architecture:Transformer Open Weights Cold

Elliot4AI/Dugong-Llama2-7b-chinese is a 7 billion parameter Llama-2-based causal language model fine-tuned for Chinese language processing. Developed by Elliot4AI, this model leverages a Chinese dataset for supervised fine-tuning (SFT), enabling it to understand and generate responses in Chinese. It utilizes 8-bit quantization and PEFT-Lora for efficient fine-tuning, making it suitable for Chinese question-answering and conversational tasks.

Loading preview...

Overview

Elliot4AI/Dugong-Llama2-7b-chinese is a 7-billion parameter language model built upon the Llama-2-7b-hf architecture. It has undergone supervised fine-tuning (SFT) specifically with a Chinese dataset, Elliot4AI/openassistant-guanaco-chinese, to enhance its capabilities in the Chinese language.

Key Characteristics

  • Base Model: Llama-2-7b-hf
  • Parameter Count: 7 billion
  • Language Focus: Primarily Chinese, enabling question-answering and conversational interactions in Chinese.
  • Fine-tuning Method: Utilizes PEFT-Lora for efficient adaptation.
  • Efficiency: Implements 8-bit quantization during the fine-tuning process.

Use Cases

This model is particularly well-suited for applications requiring:

  • Chinese Language Understanding: Processing and interpreting Chinese text.
  • Chinese Question Answering: Generating relevant answers to questions posed in Chinese.
  • Chinese Conversation: Engaging in dialogue and generating coherent responses in Chinese.