ByteDance-Seed publishes AHN-Mamba2 finetune of Qwen 2.5 Instruct 3B
ByteDance Seed published AHN-Mamba2, a finetuned variant of Qwen 2.5 Instruct 3B using the Mamba2 architecture, on Hugging Face under an other license. The repository references arxiv:2510.07318. No capability or benchmark data is disclosed in the source.
PUBLISHED2025-10-08
OBSERVED2026-08-11
AGE10mo
SOURCES1
- Published on Hugging Face under
ByteDance-Seed/AHN-Mamba2-for-Qwen-2.5-Instruct-3B - Pipeline: text-generation, library: transformers
- Base model: Qwen/Qwen2.5-3B
- License: other
- Repository created 2025-10-08, last modified 2025-10-24
- Associated paper: arxiv:2510.07318
COMMUNITY
No curated reactions recorded for this event. Facts and takes are kept in separate layers — community context is added by hand, never blended into the record above.