diff --git a/README.md b/README.md index 9d5e127a750fdd3fd49490017aab0436cffc31b7..78b4362c37ff1661f8bed9434554920a6b836757 100644 --- a/README.md +++ b/README.md @@ -3,52 +3,94 @@ license: mit task_categories: - other tags: +- smoldataenvs - rl-environment - agent - data-analysis -- benchmark +- reinforcement-learning - code-agent - harbor - openenv --- -[](https://huggingface.co/spaces/HuggingFaceH4/harbor-visualiser?dataset=FineEnvs/data-agent-harbor-test) -# 🎯 Data Agent — Harbor (test) +
-Packaged in [**Harbor**](https://github.com/huggingface/OpenEnv) format, ready to run.
+# 📊 SmolDataEnvs — Harbor (test)
-## Where it comes from
-Built from the [**jupyter-agent dataset**](https://huggingface.co/datasets/jupyter-agent/jupyter-agent-dataset)
-(real data-science notebooks over Kaggle datasets). Every question–answer pair was extracted and
-then **verified** — strong agent models must reproduce the gold answer in a live sandbox under
-deterministic grading — so **each task is known-solvable and unambiguously gradable.** This split
-is held out from training.
+**5.5K+ RL tasks for hill-climbing small models in code and data science.**
+
+[](https://huggingface.co/collections/FineEnvs/smoldataenvs)
+[](https://huggingface.co/spaces/HuggingFaceH4/harbor-visualiser?dataset=FineEnvs/SmolDataEnvs-harbor-test)
+
+