Behavioral Lift: Annotations of Reasoning Behaviors Across 15 AI Models

15,282 behavioral annotations of LLM and VLM reasoning traces were collected by neulab. The dataset covers responses from 15 models across 6 benchmarks, with each row containing correctness and a JSON-encoded behavioral annotation. It was last updated on 2026-05-08.

Use Cases

• Identify reasoning behaviors correlated with model correctness based on the behavioral annotations.
• Benchmark the reasoning performance of different AI models across multiple tasks.
• Analyze failure modes in model reasoning traces to guide future model development.

Strengths

• Contains 15,282 annotated model responses, providing a substantial sample for analysis.
• Covers 15 different AI models and 6 benchmarks, enabling comparative studies.
• Includes JSON-encoded behavioral annotations, which likely contain detailed reasoning behavior labels.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is known, but the specific file formats and data size are unknown.
• Freshness should be verified as the last update was 2026-05-08.

Provenance

Source

neulab via Hugging Face

Collection Method

Annotations of model reasoning traces from benchmarks.

Freshness

Last updated 2026-05-08 15:01:45.