Hung-Yueh Chiang
@hychiang.bsky.social
Ph.D. student @ UT
https://hychiang.info/
https://hychiang.info/
Excited to share that our #ICML2025 paper “Quamba2: A Robust and Scalable Post-training Quantization Framework for Selective State Space Models”!
July 17, 2025 at 5:10 PM
Excited to share that our #ICML2025 paper “Quamba2: A Robust and Scalable Post-training Quantization Framework for Selective State Space Models”!
We’re excited to pre-release our latest work: Quamba2
🔧 Supports W4A8 / W4A16 / W4AX / W8A8 for Mamba1 and Mamba2
🚀 Achieves 4× memory reduction and 3× generation speedup
⚡️ Enables 8B model inference on Orin Nano 8G at 13 tokens/sec
🔥 Outperforms W4A8KV4 Llama3-8B in both speed and quality
🔧 Supports W4A8 / W4A16 / W4AX / W8A8 for Mamba1 and Mamba2
🚀 Achieves 4× memory reduction and 3× generation speedup
⚡️ Enables 8B model inference on Orin Nano 8G at 13 tokens/sec
🔥 Outperforms W4A8KV4 Llama3-8B in both speed and quality
April 5, 2025 at 5:27 PM
We’re excited to pre-release our latest work: Quamba2
🔧 Supports W4A8 / W4A16 / W4AX / W8A8 for Mamba1 and Mamba2
🚀 Achieves 4× memory reduction and 3× generation speedup
⚡️ Enables 8B model inference on Orin Nano 8G at 13 tokens/sec
🔥 Outperforms W4A8KV4 Llama3-8B in both speed and quality
🔧 Supports W4A8 / W4A16 / W4AX / W8A8 for Mamba1 and Mamba2
🚀 Achieves 4× memory reduction and 3× generation speedup
⚡️ Enables 8B model inference on Orin Nano 8G at 13 tokens/sec
🔥 Outperforms W4A8KV4 Llama3-8B in both speed and quality