Effective Distillation to Hybrid xLSTM Architectures
Paper • 2603.15590 • Published • 31
Korbinian Poeppel
korbip
AI & ML interests
LLMs, Hardware-Efficient Models, Recurrent Sequence Modeling
Recent Activity
upvoted a paper 1 day ago
Effective Distillation to Hybrid xLSTM Architectures upvoted a paper 7 months ago
pLSTM: parallelizable Linear Source Transition Mark networks updated a dataset 9 months ago
ml-jku/arrow_pointing_extrapolationOrganizations
upvoted a paper 1 day ago
upvoted a paper 7 months ago
updated a
dataset 9 months ago
published a
dataset 9 months ago
authored 6
papers 9 months ago
xLSTM: Extended Long Short-Term Memory
Paper • 2405.04517 • Published • 18
A Large Recurrent Action Model: xLSTM enables Fast Inference for Robotics Tasks
Paper • 2410.22391 • Published • 22
FlashRNN: Optimizing Traditional RNNs on Modern Hardware
Paper • 2412.07752 • Published
Vision-LSTM: xLSTM as Generic Vision Backbone
Paper • 2406.04303 • Published • 1
xLSTM 7B: A Recurrent LLM for Fast and Efficient Inference
Paper • 2503.13427 • Published • 3
pLSTM: parallelizable Linear Source Transition Mark networks
Paper • 2506.11997 • Published • 10
upvoted an article 10 months ago
Article
xLSTM-based time series model TiRex significantly outperforms competing models in forecasting accuracy
•
12
updated a
model about 1 year ago
pip install transformers fork: Permission denied (publickey)
3
#8 opened about 1 year ago
by 
gue22
What is the ratio of mLSTM to sLSTM blocks & nature of the recurrence of sLSTM
2
#7 opened about 1 year ago
by 
David341
Set triton kernels as default. Remove deprecated config fields.
#5 opened about 1 year ago
by 
maxmbeck
Update readme with instructions on how to change the kernels.
#6 opened about 1 year ago
by maxmbeck
liked a
model over 1 year ago
Update LICENSE to clarify formulation
#3 opened over 1 year ago
by 
JulianFeinauer
Run it on non-Nvidia hardware?
3
#1 opened over 1 year ago
by David341