Article
CommunityJev Models & Reimplementations
TimesJev
arXiv paper proposing TimesJev, a time series System 1 model in Jev's instant-decision shape. It formalizes System 1 sufficiency and learns a compressed representation preserving key frequency patterns, reporting over 62% better results and 62.9% less inference time than fine-tuned reasoning models.
- Format
- Article
- Reading time
- 2 min
- Published by
- Community
- Added
- Nov 2026
Highlights
- Proposes TimesJev, a time series System 1 model that decides instantly instead of using long chains of thought.
- Formalizes System 1 sufficiency in time series reasoning to motivate a compressed representation.
- Compression objective combines a sufficiency divergence and a realizability divergence.
- Reports outperforming reasoning models by over 62% across 13 datasets.
- Reports 62.9% less inference time on average than fine-tuned reasoning models.
Questions about this entry
Short answers drawn from the linked project or page.
TimesJev is a time series System 1 model proposed in an arXiv paper. The authors describe it as a very early attempt to bring the instant-decision shape of models such as Jev to time series reasoning, rather than relying on long chains of thought that consume thousands of tokens.
The paper states that raw time series are not compatible with the embedding space of System 1 models, which hinders reasoning performance, especially when an input series contains thousands of numeric points. TimesJev addresses this gap with a compressed representation.
It learns a time series compression with a gating mechanism for sufficiency, producing a compressed representation that preserves key frequency patterns. The selected representation is then aligned with the System 1 model's embedding space for realizability.
The paper derives a dimension reduction objective combining two terms: a sufficiency divergence that measures the core information lost during compression, and a realizability divergence that quantifies how incompatible the System 1 model is with the optimal prediction given that representation.
Experiments on 13 datasets report that TimesJev outperforms reasoning models by over 62% while using 62.9% less inference time on average than their fine-tuned versions. The comparison is against fine-tuned reasoning models, not against untuned baselines.
The paper reports experiments on 13 datasets. Those experiments cover both the accuracy comparison against reasoning models and the average inference-time comparison against their fine-tuned versions.
The authors describe it as a very early attempt at a time series System 1 model, and the reported gains are averages over 13 datasets. The digest does not state a code release, license, or deployment surface for TimesJev.
More Jev resources like this
Zuse is a family of three decision models (4B, 9B, 27B) that train a low-rank adapter and pointer head on frozen open-weight bases for 27 European languages. The 27B model reaches 87.5% accuracy with pooled expected calibration error of 0.016 on an 11-source European suite.
Jev Models & ReimplementationsArticle#arxiv#open-models#calibration
Community
arXiv paper introducing ReListen, a training-free method in which a frozen audio language model runs a Jev-style scan: it scores each 20-second window by the probability it helps answer a question, then answers from the highest-scoring minute of hour-long recordings.
Jev Models & ReimplementationsArticle#arxiv#research#benchmark
Community
Open-weight vision-language judge that takes a task instruction plus 10 frames of a robot manipulation attempt and predicts success or failure with calibrated probabilities in a single forward pass. Released with inference code and a 20,628-prediction evaluation set.
Jev Models & ReimplementationsArticle#arxiv#open-models#evaluation
Community
arXiv paper proposing JevRec, which encodes recommendation items as variable-length semantic identifiers and adds a Jev-like decision modeling component with separate heads for length prediction and content generation, evaluated on three public datasets.
Jev Models & ReimplementationsArticle#arxiv#research#benchmark
Community
arXiv paper on a small candidate formalizer trained on offline teacher-generated lists that builds explicit, validated alternatives from task context for a frozen decision model, evaluated on CALVIN-D episodes and annotations.
Jev Models & ReimplementationsArticle#arxiv#research#benchmark
Community
A family of 0.8B-9B ophthalmic decision models that score candidate answers without autoregressive decoding, trained on RetiJev-Data: 78,158 clinical-decision and diagnosis questions and 50,074 guideline-grading questions. RetiJev-9B answers a case in 158 ms on one GPU.
Jev Models & ReimplementationsArticle#arxiv#open-models#benchmark
Community
From the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.