kairawal/Llama-3.2-3B-Instruct-ZH-SynthDolly-r16alpha32-E1-S73

TEXT GENERATIONConcurrency Cost:1Model Size:3.2BQuant:BF16Ctx Length:32kTool Calling:SupportedPublished:May 7, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

The kairawal/Llama-3.2-3B-Instruct-ZH-SynthDolly-r16alpha32-E1-S73 is a 3.2 billion parameter instruction-tuned Llama model, finetuned by kairawal from unsloth/llama-3.2-3b-Instruct. This model was trained using Unsloth and Huggingface's TRL library, enabling faster training. With a 32768 token context length, it is designed for instruction-following tasks, particularly in Chinese, as indicated by the 'ZH' in its name.

Loading preview...

Model Overview

The kairawal/Llama-3.2-3B-Instruct-ZH-SynthDolly-r16alpha32-E1-S73 is a 3.2 billion parameter instruction-tuned language model developed by kairawal. It is finetuned from the unsloth/llama-3.2-3b-Instruct base model, leveraging the Unsloth library for accelerated training, which reportedly made the training process 2x faster. The model also utilized Huggingface's TRL library during its finetuning.

Key Characteristics

  • Base Model: Finetuned from unsloth/llama-3.2-3b-Instruct.
  • Parameter Count: 3.2 billion parameters.
  • Context Length: Supports a substantial context window of 32768 tokens.
  • Training Efficiency: Benefits from Unsloth's optimizations for faster training.
  • Instruction Following: Designed for instruction-based tasks, indicated by "Instruct" in its name.
  • Multilingual Focus: The "ZH" in the model name suggests a specialization or strong performance in Chinese language tasks.

Intended Use Cases

This model is suitable for developers looking for a compact yet capable instruction-following model, especially for applications requiring a large context window and potentially strong performance in Chinese. Its efficient training methodology makes it an interesting option for those prioritizing faster iteration cycles.