AgentTrove
- Type
- dataset
- Venue
- open-thoughts
- Year
- 2026
- Source
- huggingface
- Access
- free
- Language
- en
- Added
- 2026-07-17T20:02:07.913649+00:00
- Verified
- 2026-07-17T20:02:07.913649+00:00
Summary
At 1.7 million rows, AgentTrove is **4× the size of the [Nemotron Terminal Corpus](https://huggingface.co/datasets/laion/nemotron-terminal-corpus-unified)** (430 K rows), the previous largest open-source agentic trace dataset.
Keywords
hf-dataset language-modeling · -code · -reinforcement-learning parquet text datasets dask polars mlcroissant agent code agentic-traces reinforcement-learning terminus-2 harbor agent-traces
Topics
Language Modeling, Code, Reinforcement Learning
Research notes
- downloads=2191; likes=188