HallusionBench
PulseAugur coverage of HallusionBench — every cluster mentioning HallusionBench across labs, papers, and developer communities, ranked by signal.
3 day(s) with sentiment data
-
New framework tackles visual hallucinations in multimodal LLMs · 2 sources tracked
Researchers have introduced DreamRealm, a new framework designed to combat visual hallucinations in multimodal large language models (MLLMs). This framework works by identifying and materializing latent hallucinated res…
-
New MACR framework enables dynamic relay thinking for multi-agent AI
Researchers have introduced Multi-Agent Confidence-guided Relay (MACR), a novel framework for multi-agent collaboration that shifts from group thinking to a relay-based approach. In MACR, reasoning ownership is dynamica…
-
New RECAP method boosts vision-language model reasoning
Researchers have developed a new training method called RECAP to improve vision-language models (VLMs) by addressing temporal grounding failures. This method enhances how models utilize image evidence during multi-step …
-
LoopLook enhances Vision-Language Models with iterative visual perception
Researchers have developed LoopLook, a novel recurrent perception framework designed to enhance Vision-Language Models (VLMs). This system allows VLMs to iteratively refine visual evidence in response to specific questi…
-
New HalluPrism tool diagnoses MLLM failures with improved accuracy
Researchers have developed HalluPrism, a new diagnostic tool designed to better understand the failure modes of Multimodal Large Language Models (MLLMs). This method involves re-running model answers after introducing v…
-
VLMs' reasoning chains offer better uncertainty signals than answer entropy, study finds
A new research paper explores the effectiveness of "thinking chains" in visual language models (VLMs) for quantifying uncertainty. The study found that while some models like Qwen3-VL-8B-Thinking exhibit a complete coll…
-
New methods tackle hallucinations in video large multimodal models
Researchers have developed several new methods to combat hallucinations in video large multimodal models (VLMMs). One approach, MultiToP, refines unreliable visual tokens before language generation by selectively substi…
-
Visual Para-Thinker introduces parallel reasoning to multimodal LLMs
Researchers have introduced Visual Para-Thinker, a novel framework for parallel reasoning in multimodal large language models (MLLMs). This approach shifts from vertical scaling of reasoning depth to a parallel strategy…
-
New theory guides LLM action decisions by selecting optimal controller classes
Researchers have introduced a "Regime Theory" to guide how large language models decide on the best action for a given input. The theory categorizes controllers into four classes, from simple fixed actions to complex pr…