Sign in to view source links and access this dataset
Description
2,238 tasks form a native training suite for the data-agent post-training pipeline. The dataset, sourced from AdithyaSK/data_agent_rl_environment_train, includes task questions, configurations, and a Docker environment. It was last updated on July 10, 2026.
Use Cases
Training reinforcement learning agents based on the 2,238 provided tasks.
Benchmarking agent performance using the task-local verifier system mentioned in the description.
Developing post-training pipeline components within the described Docker environment.
Strengths
Contains 2,238 distinct training tasks.
Includes a defined Docker environment for consistent execution.
Provides a task-local verifier system for reward calculation.
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
Row count is unknown, which may limit suitability assessment.
Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
Source
AdithyaSK/data_agent_rl_environment_train on Hugging Face, converted by BenchFlow.
Collection Method
Native package entrypoint is task.md; likely programmatically generated or assembled for training.
Freshness
Last updated 2026-07-10 21:14:45; freshness should be verified.
License is Apache-2.0. Requires Docker to utilize the provided environment.