Machine Learning

LLM Research

New Method Detects LLM Hallucinations via Internal State Analysis

Researchers propose a novel framework for visualizing and benchmarking factual hallucinations in large language models by analyzing internal neural activations and clustering patterns.

Machine Learning

Data Leakage in ML: Why Train-Test Splits Must Come First

Data leakage silently destroys model validity. Learn why preprocessing before splitting contaminates your test set and how to build pipelines that preserve true model performance.

synthetic data

PRISM: Structure-Aware Privacy Budgets for Synthetic Data

New research introduces PRISM, a differentially private synthetic data framework using structure-aware budget allocation to optimize prediction accuracy while maintaining privacy guarantees.

LLM evaluation

MILE-RefHumEval: Multi-LLM Framework for Human-Aligned AI Evaluat

New research introduces a reference-free evaluation framework using multiple independent LLMs to assess AI outputs with better human alignment than single-judge approaches.

LLM Agents

PABU: Making LLM Agents Smarter Through Progress-Aware Updates

New research introduces PABU, a framework that helps LLM agents track their progress and update beliefs more efficiently, reducing computational waste in multi-step reasoning tasks.

LLM

LLM Parameters Explained: Weights, Biases, and Scale Demystified

Understanding LLM parameters is key to grasping how AI models generate text, images, and video. Learn what weights and biases actually do and why model scale matters.

prompt engineering

7 Prompt Engineering Methods That Boost LLM Output Quality

From chain-of-thought reasoning to self-consistency sampling, these seven prompt engineering techniques can dramatically improve how large language models respond to complex queries.

LLM evaluation

LLM Judges Exposed: Research Reveals Hidden Evaluation Shortcuts

New research uncovers systematic shortcuts in LLM-based evaluation systems, revealing how AI judges may rely on superficial patterns rather than genuine quality assessment.

Hugging Face

Hugging Face Transformers v5: Simplified APIs for AI Development

Hugging Face releases Transformers v5 with cleaner APIs, unified model loading, and breaking changes that simplify building AI applications across text, image, and video domains.

LLM Research

Sketch-and-Walk: New Sparse Attention Method Speeds Up LLM Infere

Researchers propose a two-phase sparse attention mechanism that scouts relevant tokens before full computation, promising significant efficiency gains for large language model inference.

AI research

AIRS-Bench: New Benchmark Suite Tests AI Research Agents

A new benchmark suite evaluates how well AI agents can perform frontier research tasks, measuring capabilities from literature review to hypothesis generation and experimental design.

Transformers

Positional Encoding Methods: Why Token Order Matters in AI

Transformers process tokens in parallel, losing sequence information. Four positional encoding methods—sinusoidal, learned, RoPE, and ALiBi—solve this fundamental challenge differently.