Toward Generalizable Evaluation in the LLM Era: A Survey Beyond Benchmarks
read the original abstract
Large Language Models (LLMs) are advancing at an amazing speed and have become indispensable across academia, industry, and daily applications. To keep pace with the status quo, this survey probes the core challenges that the rise of LLMs poses for evaluation. We identify and analyze two pivotal transitions: (i) from task-specific to capability-based evaluation, which reorganizes benchmarks around core competencies such as knowledge, reasoning, instruction following, multi-modal understanding, and safety; and (ii) from manual to automated evaluation, encompassing dynamic dataset curation and "LLM-as-a-judge" scoring. Yet, even with these transitions, a crucial obstacle persists: the evaluation generalization issue. Bounded test sets cannot scale alongside models whose abilities grow seemingly without limit. We will dissect this issue, along with the core challenges of the above two transitions, from the perspectives of methods, datasets, evaluators, and metrics. Due to the fast evolving of this field, we will maintain a living GitHub repository (links are in each section) to crowd-source updates and corrections, and warmly invite contributors and collaborators.
This paper has not been read by Pith yet.
Forward citations
Cited by 9 Pith papers
-
Security in LLM-as-a-Judge: A Comprehensive SoK
The first SoK on LLM-as-a-Judge security organizes attacks targeting judges, attacks using judges, defenses leveraging judges, and security-domain applications while flagging vulnerabilities.
-
Evaluating LLMs on Large-Scale Graph Property Estimation via Random Walks
EstGraph benchmark evaluates LLMs on estimating properties of very large graphs from random-walk samples that fit in context limits.
-
OpenSkillEval: Automatically Auditing the Open Skill Ecosystem for LLM Agents
OpenSkillEval automatically builds realistic tasks from evolving artifacts to audit skill effectiveness in LLM agents, finding that skill use depends on model and framework and that many popular skills do not outperfo...
-
OpenSkillEval: Automatically Auditing the Open Skill Ecosystem for LLM Agents
OpenSkillEval dynamically builds task instances across five application domains to evaluate 30 open skills with over 600 tests, finding that skill use depends heavily on model and framework and that many popular skill...
-
Hint Tuning: Less Data Makes Better Reasoners
Hint Tuning uses an instruct model as a difficulty probe to create 1K multi-level hint examples that train reasoning models to calibrate chain-of-thought length, cutting tokens by 31.5% on average across 4B-32B models...
-
Hint Tuning: Less Data Makes Better Reasoners
Hint Tuning reduces token usage 24-66% (31.5% avg) in reasoning models via 1K self-annotated samples aligned to an instruct model's capabilities while keeping benchmark accuracy.
-
Safe for Whom? Rethinking How We Evaluate the Safety of LLMs for Real Users
LLM safety evaluations for personal advice must test responses against diverse user vulnerability profiles, since context-blind ratings overestimate safety and realistic prompt context does not fix the problem.
-
Reasoning emerges from constrained inference manifolds in large language models
Reasoning in LLMs emerges from inference dynamics forming constrained low-dimensional manifolds that preserve non-degenerate information volume, rather than from compression alone.
-
Discriminatory Compliance: How LLMs Answer Queries from Protected Groups
State-of-the-art LLMs respond inconsistently to queries from protected-group personas, with some responses omitting key information that should be provided.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.