Andreas Auer (@AndAuer) profile picture

Science @ AWS | Time Series - Foundation Models | 🏠 based in Berlin

Berlin
Joined May 2022
Photo by Andreas Auer (@AndAuer)Photo by Andreas Auer (@AndAuer)
We’re excited to introduce TiRex β€” a pre-trained time series forecasting model based on an xLSTM architecture.
5
22
6
69
14,925
Andreas Auer retweeted
πŸš€ Excited to share our new paper on scaling laws for xLSTMs vs. Transformers. Key result: xLSTM models Pareto-dominate Transformers in cross-entropy loss. - At fixed FLOP budgets β†’ xLSTMs perform better - At fixed validation loss β†’ xLSTMs need fewer FLOPs 🧡 Details in thread
14
40
13
231
85,123
Ever wondered how linear RNNs like #mLSTM (#xLSTM) or #Mamba can be extended to multiple dimensions? Check out "pLSTM: parallelizable Linear Source Transition Mark networks". #pLSTM works on sequences, images, (directed acyclic) graphs. Paper link: arxiv.org/abs/2506.11997
4
42
2
134
15,164
We also introduce Contiguous Patch Masking, a masking strategy for training that facilitates TiRex’s ability to generate coherent quantile estimates in long-horizon settings, without sacrificing short-term accuracy.
1
2
410
Check out the paper and the model here: πŸ“ Paper: arxiv.org/abs/2505.23719 πŸ€— Huggingface: huggingface.co/NX-AI/TiRex πŸ§‘β€πŸ’» Github: github.com/NX-AI/tirex Joint work with Patrick Podest, @ido87 , Sebastian BΓΆck, @gklambauer , @HochreiterSepp , and the great team of @nx_ai_com .
1
12
411
Introducing Vision-LSTM - making xLSTM read images 🧠It works ... pretty, pretty well πŸš€πŸš€ But convince yourself :) We are happy to share code already! πŸ“œ: arxiv.org/abs/2406.04303 πŸ–₯️: nx-ai.github.io/vision-lstm/ All credits to my stellar PhD @benediktalkin
5
76
11
380
78,063
Andreas Auer retweeted
Really excited to announce Energy-based Hopfield Boosting. πŸŽ‰ Hopfield Boosting advances the state-of-the-art in OOD detection. πŸš€ Our new energy function allows a model to focus on hard training examples close to the decision boundary between in-distribution and OOD samples. 🧡
1
40
5
151
25,318
Andreas Auer retweeted
The #xLSTM is finally live! What an exciting day! How far do we get in language modeling with the LSTM compared to State-of-the-Art LLMs? I would say pretty, pretty far! How? We extend the LSTM with Exponential Gating and parallelizable Matrix Memory! arxiv.org/pdf/2405.04517
7
37
2
151
18,656
I am so excited that xLSTM is out. LSTM is close to my heart - for more than 30 years now. With xLSTM we close the gap to existing state-of-the-art LLMs. With NXAI we have started to build our own European LLMs. I am very proud of my team. arxiv.org/abs/2405.04517
45
358
76
1,819
276,755
xLSTM is out -- putting LSTM networks on steroids to become a more than serious LLM competitor. How? Via exponential gating and enhanced (cell state) memory capacities. Does it work? Oh, yeah πŸš€πŸš€ arxiv.org/abs/2405.04517
6
72
7
353
40,612
Interested in a semantic memory for reinforcement learning? I was recently invited to a podcast talking about our #NeurIPS2023 paper: Semantic HELM (arxiv.org/abs/2306.09312). In case you are interested, you can stream the episode here: open.spotify.com/episode/4n2…
Excited to share our latest work on a semantic and interpretable memory module for RL! Complementary to recent developments in the realm of explainable AI, we focus on interpretability w.r.t. the memory of an agent. 1/n
1
15
23
2,051
Andreas Auer retweeted
Our latest paper challenges the status quo in Unsupervised Learning for Combinatorial Optimization (CO) problems.πŸ’‘ πŸ”„ Unveiling the limitations of popular Mean Field methods, we showcase the power of Autoregressive Approximations of solving CO Problems on graphs. 🧡
1
25
2
53
13,521
Time series come with tradition at @HochreiterSepp's institute - @AndAuer puts a new perspective on conformal predictions for time series data - great work to appear @NeurIPSConf !
Interested in reliable prediction intervals for time series? πŸš€ Excited to announce our latest work, which proposes HopCPT, was accepted at #Neurips2023 πŸŽ‰. HopCPT tackles the challenges of time series data for Conformal Prediction with the help of Modern Hopfield Networks. 1/4
2
27
2,832
Super paper on reliable prediction for time series at NeurIPS. Conformal Predictions adapted to time series yields excellent results.
Interested in reliable prediction intervals for time series? πŸš€ Excited to announce our latest work, which proposes HopCPT, was accepted at #Neurips2023 πŸŽ‰. HopCPT tackles the challenges of time series data for Conformal Prediction with the help of Modern Hopfield Networks. 1/4
14
63
14,643
Interested in reliable prediction intervals for time series? πŸš€ Excited to announce our latest work, which proposes HopCPT, was accepted at #Neurips2023 πŸŽ‰. HopCPT tackles the challenges of time series data for Conformal Prediction with the help of Modern Hopfield Networks. 1/4
4
42
5
156
42,450
Additionally, the updatable Hopfield Memory allows HopCPT to mitigate shifts in the error distribution πŸ”„. The empirical evaluation on 6 real-world datasets shows that HopCPT establishes a new state of the art in Conformal Prediction for time series. πŸ“ˆ 3/4
1
6
983