FuseAI/FuseChat-Llama-3.2-3B-Instruct
TEXT GENERATIONConcurrency Cost:1Model Size:3.2BQuant:BF16Ctx Length:32kPublished:Dec 6, 2024Architecture:Transformer0.0K Loading

FuseAI/FuseChat-Llama-3.2-3B-Instruct is a 3 billion parameter instruction-tuned language model developed by FuseAI, part of the FuseChat-3.0 series. This model is created through an implicit model fusion process, leveraging Supervised Fine-Tuning (SFT) and Direct Preference Optimization (DPO) to integrate capabilities from larger source LLMs like Gemma-2-27B-It, Mistral-Large-Instruct-2407, Qwen-2.5-72B-Instruct, and Llama-3.1-70B-Instruct into a more compact Llama-3.2-3B-Instruct base. It excels in general conversation, instruction following, mathematics, and coding tasks, demonstrating an average performance improvement of 5.0 points over the base Llama-3.2-3B-Instruct across 14 benchmarks.

Loading preview...