IST Austria Distributed Algorithms and Systems Lab

university

https://ist.ac.at/en/research/alistarh-group/

https://github.com/IST-DASLab

AI & ML interests

None defined yet.

Recent Activity

mkleinegger authored a paper about 20 hours ago

Disaggregated Quantization: Specializing LLM Prefill and Decode

mkleinegger authored a paper about 20 hours ago

One Simple Trick for Improving the Performance of Energy-Limited Local Inference and Training

softmax submitted a paper 1 day ago

WUSH-KV: KV Cache Quantization with Data-Adaptive Transforms

View all activity

Papers

WUSH-KV: KV Cache Quantization with Data-Adaptive Transforms

Disaggregated Quantization: Specializing LLM Prefill and Decode

View all Papers

authored 2 papers about 20 hours ago

Disaggregated Quantization: Specializing LLM Prefill and Decode

Paper • 2609.26333 • Published 10 days ago • 91

One Simple Trick for Improving the Performance of Energy-Limited Local Inference and Training

Paper • 2609.11936 • Published Jul 2

submitted a paper to Daily Papers 1 day ago

WUSH-KV: KV Cache Quantization with Data-Adaptive Transforms

Paper • 2609.38121 • Published 3 days ago • 14

authored a paper 2 days ago

WUSH-KV: KV Cache Quantization with Data-Adaptive Transforms

Paper • 2609.38121 • Published 3 days ago • 14

updated 2 models 2 days ago

ISTA-DASLab/Qwen3-8B-MatGPTQ

8B • Updated 2 days ago • 32

ISTA-DASLab/Llama-3.1-8B-Instruct-MatGPTQ

8B • Updated 2 days ago • 102

updated 2 models 3 days ago

ISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RCO-Coder-GGUF

Image-Text-to-Text • 117B • Updated 3 days ago • 261k • 187

ISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RCO-GGUF

Image-Text-to-Text • 177B • Updated 3 days ago • 1.14M • 438

updated a collection 3 days ago

MoESQ

Paired-4:8 + NVFP4 W4A4 expert compressed MoE models for NVIDIA Blackwell Sparse Tensor Cores. • 3 items • Updated 3 days ago • 1

updated a model 3 days ago

ISTA-DASLab/Qwen3.5-397B-A17B-P48NVFP4-MoESQ

Image-Text-to-Text • 258B • Updated 3 days ago • 26

published a model 3 days ago

ISTA-DASLab/Qwen3.5-397B-A17B-P48NVFP4-MoESQ

Image-Text-to-Text • 258B • Updated 3 days ago • 26

updated a collection 3 days ago

MoESQ

Paired-4:8 + NVFP4 W4A4 expert compressed MoE models for NVIDIA Blackwell Sparse Tensor Cores. • 3 items • Updated 3 days ago • 1

published a model 3 days ago

ISTA-DASLab/Qwen3-30B-A3B-P48NVFP4-MoESQ

Text Generation • 20B • Updated 3 days ago • 114 • 1

updated a model 3 days ago

ISTA-DASLab/Qwen3-30B-A3B-P48NVFP4-MoESQ

Text Generation • 20B • Updated 3 days ago • 114 • 1

updated a collection 3 days ago

MoESQ

Paired-4:8 + NVFP4 W4A4 expert compressed MoE models for NVIDIA Blackwell Sparse Tensor Cores. • 3 items • Updated 3 days ago • 1

updated a model 3 days ago

ISTA-DASLab/Kimi-K2.5-P48NVFP4-MoESQ

Text Generation • 646B • Updated 3 days ago • 115

in ISTA-DASLab/Qwen3.8-27B-NVFP4-prefiller 3 days ago

Add pipeline tag

#3 opened 3 days ago by

published a model 3 days ago

ISTA-DASLab/Kimi-K2.5-P48NVFP4-MoESQ

Text Generation • 646B • Updated 3 days ago • 115

authored 2 papers 3 days ago

Disaggregated Quantization: Specializing LLM Prefill and Decode

Paper • 2609.26333 • Published 10 days ago • 91

Apertus LLM Family Expansion via Distillation and Quantization

Paper • 2605.29128 • Published May 27