Josef Liyanjun Chen
PulseAugur coverage of Josef Liyanjun Chen — every cluster mentioning Josef Liyanjun Chen across labs, papers, and developer communities, ranked by signal.
2 day(s) with sentiment data
-
Research paper questions effectiveness of excessive planning in AI world models
A new research paper titled "When More Planning Hurts: Learned Planners for World Models" explores the effectiveness of learned planners in world models for robotics. The study, authored by Josef Liyanjun Chen and other…
-
RoboJev-27B: Open Vision-Language Model for Robot Task Success Detection
Researchers have developed RoboJev-27B, an open-weight vision-language model designed to determine if a robot has successfully completed a manipulation task from video data. This model, a variant of JEV-27B-VL, was fine…
-
FlavourBench benchmark ranks language models using culinary tasks · 2 sources tracked
A new benchmark called FlavourBench has been introduced to evaluate frontier language models using a culinary system for executable ground truth. This system, named Epicure, provides dense, executable scores for tasks i…
-
New research reveals co-failure ceiling limits LLM ensemble gains
A new research paper introduces the concept of a "co-failure ceiling" to explain the limitations of combining multiple large language models. The study demonstrates that the accuracy gains from ensemble methods like rou…
-
AURA-Mem cuts robot AI memory writes with action-gated system
Researchers have developed AURA-Mem, a novel memory system designed for embodied AI agents operating on resource-constrained edge hardware. Unlike datacenter-focused KV-caches, AURA-Mem uses a constant-size recurrent me…
-
AI inference latency limited by more than memory bandwidth, study finds
A new paper reveals that the inference performance of physical AI systems, such as robots and autonomous vehicles, is not solely limited by memory bandwidth as previously assumed. The research demonstrates that while ba…
-
New embeddings map food ingredient relationships using recipes and chemistry
Researchers have developed "Epicure," a set of three skip-gram embeddings trained on a large multilingual recipe corpus. These embeddings are designed to capture the relationships between food ingredients, considering b…
-
AI decodes culinary intuition, mapping flavor dimensions in ingredient embeddings
Researchers have developed "Epicure," a system that extracts multidimensional flavor structures from food ingredient embeddings. By augmenting existing embeddings with an LLM-driven curation pipeline, they consolidated …