kairawal/Llama-3.1-8B-Instruct-ZH-SynthDolly-r16alpha32-E1-S73

TEXT GENERATIONConcurrency Cost:1Model Size:8BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:May 12, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

The kairawal/Llama-3.1-8B-Instruct-ZH-SynthDolly-r16alpha32-E1-S73 is an 8 billion parameter instruction-tuned language model, finetuned by kairawal from Meta's Llama-3.1-8B-Instruct. This model was trained using Unsloth and Huggingface's TRL library, indicating an optimization for efficient training. Its primary differentiation lies in its specific finetuning, likely for specialized instruction-following tasks, potentially involving Chinese language (ZH) and synthetic data (SynthDolly) as suggested by its name.

Loading preview...

Overview

This model, kairawal/Llama-3.1-8B-Instruct-ZH-SynthDolly-r16alpha32-E1-S73, is an 8 billion parameter instruction-tuned language model. It was developed by kairawal and is finetuned from Meta's Llama-3.1-8B-Instruct base model. The finetuning process leveraged Unsloth and Huggingface's TRL library, which are known for enabling faster and more efficient training of large language models.

Key Characteristics

  • Base Model: Finetuned from Meta-Llama-3.1-8B-Instruct, inheriting its foundational capabilities.
  • Efficient Training: Utilizes Unsloth and Huggingface TRL for optimized and accelerated finetuning.
  • Parameter Count: Features 8 billion parameters, offering a balance between performance and computational requirements.
  • License: Distributed under the Apache-2.0 license, allowing for broad use and modification.

Potential Use Cases

Given its instruction-tuned nature and the components in its name (ZH for Chinese, SynthDolly for synthetic data), this model is likely well-suited for:

  • Instruction-following tasks, particularly in a Chinese language context.
  • Applications requiring responses based on synthetic data patterns.
  • Scenarios where efficient deployment of an 8B parameter model is crucial.

Finetunes
1 models