Welcome RL Environments to the hub
An environment gives an agent a task, responds to its actions with observations, and scores the outcome.
ProofVendor claim only
Key points
- For an introduction to this interaction loop, see our blogpost on environments.
- In this release, we are focusing on the tasksets.
- An RL environment on the Hub is a dataset repo that shows up in the new RL Environments filter.
- Custom hubs, runtime registries, independent task datasets, or a GitHub list of tasks with a custom loader.
Sources (1)
- [1]Welcome RL Environments to the hubHugging Face Blog · Sep 28, 12:00 AM
An environment gives an agent a task, responds to its actions with observations, and scores the outcome.
For an introduction to this interaction loop, see our blogpost on environments.
Extractive summary: sentences quoted from the sources.
Hype check
Only the maker's wordStill only the maker's word
- Follow-up · Sep 28, 2026Add Runtime Controls to AI Agents with NVIDIA OpenShell
- Follow-up · Sep 29, 2026[AINews] Opus 5.5 is good at explainer videos
- Follow-up · Sep 30, 2026axolotl-ai-cloud/axolotl v0.20.0
- Follow-up · Sep 30, 2026Gemini 4 Argon: our next era of frontier intelligence
- Follow-up · Oct 1, 2026Build Local AI Apps with C++ and NVIDIA TensorRT RTX Samples
- Follow-up · Oct 1, 2026Build Applications on NVIDIA BlueField Faster with NVIDIA DOCA Agent Skills
Before this
- Sep 22, 2026vllm-project/vllm v0.30.0
- Aug 26, 2026huggingface/transformers v5.16.0: Release: v5.16.0
- Jul 3, 2026huggingface/transformers v5.13.0: Release v5.13.0
- Jun 10, 2026huggingface/transformers v5.11.0: Release v5.11.0
- Jun 10, 2026DiffusionGemma: 4x faster text generation
- Jun 9, 2026Introducing Gemma 4 12B: a unified, encoder-free multimodal model
