analiza aiqbench
This paper is squarely about AI: it studies parameter sharing (weight tying with looped blocks) inside the in-context learning stage of a tabular foundation model. Its results suggest that for this architecture and scale, depth matters more than having distinct parameters per layer, which could cut model size substantially while letting inference compute be tuned after training. Readers should note the comparisons are benchmark-based at a single model family and scale, so the generality of the claim is still open.
Recurrent In-Context Learning for Tabular Foundation Models
arXiv:2610.14998v1 Announce Type: cross Abstract: Tabular foundation models adapt to new datasets through in-context learning (ICL), and, like most transformers, they tie computational depth to the number of independently parameterized layers. We ask whether the ICL stage of such a model needs a distinct set of parameters for every layer, or only depth. We introduce Tab-Loop, which replaces the 12 untied ICL blocks of TabICLv2 with 4 tied blocks applied $k$ times and changes nothing else: the encoders, decoder, synthetic prior and training recipe are shared with an untied baseline trained for the same compute. At matched pretraining compute (18.2 EFLOP) and matched inference compute (12 block applications), Tab-Loop matches the baseline with 62\% fewer parameters (10.4M vs.\ 27.6M): held-out log loss is 0.4583 vs.\ 0.4588, mean accuracy on 200 TALENT datasets is 84.10\% vs.\ 83.92\%, and Elo on 38 TabArena datasets is 1409 vs.\ 1396. A 4-block untied control with Tab-Loop's parameter count, which sees 32\% more training tables for the same compute, trails both (83.46\%; Elo 1343). Because weights are shared, depth becomes a choice made at inference. Accuracy rises with the number of loops through the trained range and holds beyond it, and continued training with deeper loops extends the useful range, reaching the baseline's stage-2 scores at 4--5$\times$ its inference compute. At this scale, the ICL stage of a tabular foundation model needs depth more than it needs distinct parameters.
Abstract:Tabular foundation models adapt to new datasets through in-context learning (ICL), and, like most transformers, they tie computational depth to the number of independently parameterized layers. We ask whether the ICL stage of such a model needs a distinct set of parameters for every layer, or only depth. We introduce Tab-Loop, which replaces the 12 untied ICL blocks of TabICLv2 with 4 tied blocks applied $k$ times and changes nothing else: the encoders, decoder, synthetic prior and training recipe are shared with an untied baseline trained for the same compute. At matched pretraining compute (18.2 EFLOP) and matched inference compute (12 block applications), Tab-Loop matches the baseline with 62\% fewer parameters (10.4M vs.\ 27.6M): held-out log loss is 0.4583 vs.\ 0.4588, mean accuracy on 200 TALENT datasets is 84.10\% vs.\ 83.92\%, and Elo on 38 TabArena datasets is 1409 vs.\ 1396. A 4-block untied control with Tab-Loop's parameter count, which sees 32\% more training tables for the same compute, trails both (83.46\%; Elo 1343). Because weights are shared, depth becomes a choice made at inference. Accuracy rises with the number of loops through the trained range and holds beyond it, and continued training with deeper loops extends the useful range, reaching the baseline's stage-2 scores at 4--5$\times$ its inference compute. At this scale, the ICL stage of a tabular foundation model needs depth more than it needs distinct parameters.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2610.14998 [cs.LG]
(or arXiv:2610.14998v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.14998 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 326 cuvinte
Subiecte conexe
Build Your Own Transaction Foundation Model for Financial Intelligence
NVIDIA · 16 iun. 2026
GeneICL: A Tabular Foundation Model for Bulk Transcriptomics
arXiv · 7 oct. 2026
Adapting Linear-Time Architectures for Tabular In-Context Learning
arXiv · 30 sept. 2026
LoopICL: Looping a single transformer block to solve tabular tasks
arXiv · 30 sept. 2026