LLM evaluation benchmarks aren’t as objective as they seem. What LLM picked as the LLM as a Judge can dramatically change the outcome of the evaluation. However, the Language Model Council research suggests that the top spot on any given leaderboard might be an artifact of evaluation design rather than a reflection of superior, generalized capability.
An LLM research paper, titled “Artificial or Just Artful? explores the tension between pretraining objectives and alignment constraints in Large Language Models (LLMs). The researchers specifically investigated how models adapt their strategies when exposed to test cases from the BigCodeBench (Hard) dataset.
RAGRecon, a system to improve Cyber Threat Intelligence through the integration of Large Language Models and Retrieval-Augmented Generation.
DRAFT-RL is a evaluation framework fort LLMs designed to address critical limitations in LLM-based reasoning systems by integrating Chain-of-Draft (CoD) reasoning with multi-agent reinforcement learning.
LLM evaluation benchmarks aren’t as objective as they seem. What LLM picked as the LLM as a Judge can dramatically change the outcome of the evaluation. However, the Language Model Council research suggests that the top spot on any given leaderboard might be an artifact of evaluation design rather than a reflection of superior, generalized capability.
Humanity’s Last Exam is a multi-modal case study designed to measure the capabilities of large language models.
A new research paper from Humains-Junior language model reportedly matches the factual accuracy of GPT-4o on a specific public subset. According to the paper the Humains-Junior language model achieves this performance through an innovative method called “Exoskeleton Reasoning.”
