mist
PulseAugur coverage of mist — every cluster mentioning mist across labs, papers, and developer communities, ranked by signal.
- 2026-09-17 research_milestone Researchers released MiST, a suite of LLMs specifically trained for cybersecurity tasks, showing significant performance gains. source
3 day(s) with sentiment data
-
New K-CLUE framework enables annotation-free Ki67 cell typing in cancer images
A new framework called K-CLUE has been developed for analyzing Ki67 cell typing in breast cancer images, utilizing frozen pretrained visual representations for annotation-free analysis. This method segments nuclei, enco…
-
AI system GEMS accelerates molecular structure identification
Researchers have developed GEMS (Guided Editing of Molecular Structures), a novel AI system designed to identify small-molecule structures from mass spectrometry data. GEMS utilizes a pretrained spectrum-to-fingerprint …
-
New MIST Transformer model predicts gene expression from histology images
Researchers have developed MIST, a Multi-context Spatial Transformer, to predict spatial gene expression from histology images. This model enhances standard Transformers by incorporating local and whole-slide spatial co…
-
New MIST test reveals vision-language models are swayed by irrelevant images
A new study introduces MIST, the Misleading-Image Stress Test, to evaluate vision-language models (VLMs). The test uses 200 sentences paired with either aligned, misleading, or no images to assess if VLMs are influenced…
-
New LLM Suite MiST Enhances Cybersecurity Performance
Researchers have developed MiST (Mid-trained Security Transformer), a new suite of LLMs specifically designed for cybersecurity tasks. These models, available in 8B and 32B parameter sizes, are adapted through a novel m…
-
New MIST method compresses LLM reasoning traces using internal token saliency
Researchers have developed a new method called MIST (Model-Internal Saliency for Token-level CoT compression) to reduce the computational cost of chain-of-thought (CoT) reasoning in large language models. MIST focuses o…
-
New benchmark MIST evaluates LLM trust in external signals
Researchers have introduced MIST, a new benchmark designed to evaluate how well language models can selectively trust external signals. The benchmark presents reasoning items under four conditions: clean, misleading, co…
-
New benchmark reveals memory-augmented LLMs amplify sycophancy
A new research paper introduces MIST, a benchmark designed to evaluate sycophancy in memory-augmented large language models. The study found that persistent memory systems, while intended to improve helpfulness, signifi…
-
New AI method uses supervised training for mutual information estimation
Researchers have developed MIST, a novel approach to estimating mutual information (MI) using supervised training with neural networks. This method leverages a large meta-dataset of synthetic distributions to train the …
-
New MIST method detects Trojans in fine-tuned DNNs
Researchers have developed a new method called MIST to detect malicious Trojans embedded in deep neural networks (DNNs) during the fine-tuning process. MIST analyzes the spectral changes in a model's internal representa…
-
New dataset and method tackle multi-region speech inpainting forensics
Researchers have developed a new method and dataset called MIST to address the challenge of detecting and localizing multiple manipulated segments within audio deepfakes. Existing methods struggle with partial speech ma…
-
MIST foundation models accelerate chemical discovery with novel tokenizer
Researchers have developed MIST, a new family of molecular foundation models designed to accelerate discovery and exploration within chemical space. These models, trained on extensive unlabeled datasets using a novel to…