{"categories":[{"label":"Software Development","url":"https://skillfed.io/packages/category/software-development/17"},{"label":"Libraries","url":"https://skillfed.io/packages/category/software-development-libraries/9"},{"label":"Scientific/Engineering","url":"https://skillfed.io/packages/category/scientific-engineering/8"},{"label":"Python Modules","url":"https://skillfed.io/packages/category/software-development-libraries-python-modules/20"},{"label":"Artificial Intelligence","url":"https://skillfed.io/packages/category/scientific-engineering-artificial-intelligence/10"},{"label":"Mathematics","url":"https://skillfed.io/packages/category/scientific-engineering-mathematics/3"}],"enrichment":{"capability":"Dopamine is a research framework for prototyping reinforcement learning algorithms, providing implementations of DQN, C51, Rainbow, IQN, SAC, and PPO agents built primarily with JAX.","skillfed_tags":["reinforcement-learning","research-framework","jax-based"],"use_cases":["Prototype and test new RL algorithm ideas on Atari benchmark environments with minimal boilerplate.","Train and compare standard agents (DQN, Rainbow, PPO) to establish baselines for research papers.","Learn RL algorithm implementation by reading and modifying well-documented, battle-tested agent code.","Reproduce published results using Dopamine's implementations following Machado et al. (2018) evaluation protocols.","Extend existing agents with custom reward shaping or network architectures for domain-specific tasks."],"what_it_does":"Dopamine is a research-focused reinforcement learning framework maintained by Google that prioritizes ease of experimentation and code clarity over feature breadth. It implements six well-tested RL agents (DQN, C51, Rainbow, IQN, SAC, PPO) with active development in JAX and legacy support for TensorFlow, designed for researchers and students to prototype new ideas on Atari and Mujoco environments.\n\nThe framework emphasizes reproducibility and follows best practices from the RL literature. It ships with 19 dependencies spanning deep learning (TensorFlow, JAX, Flax), environment simulation (Gym, Gymnasium, ALE), and data handling (NumPy, Pandas, OpenCV). Installation from source is recommended for research work because the codebase is designed to be modified directly; pip installation is supported but less typical for active development.","worth_installing":"Yes, if you are conducting RL research or learning algorithm implementation. The framework is actively maintained, carries no known vulnerabilities, and provides a curated set of proven agents with clear code. The 19 dependencies and environment setup overhead make it less suitable for lightweight production inference, but ideal for experimentation and reproducible research."},"id":"dopamine-rl","links":{"html":"https://skillfed.io/packages/dopamine-rl","md":"https://skillfed.io/packages/dopamine-rl.md","pypi":"https://pypi.org/project/dopamine-rl/"},"maintenance":{"status":"active"},"meta":{"latest_release":"2024-10-31","license_spdx":null,"license_treatment":"permissive","name":"dopamine-rl","python_support":"supports_current","summary":"Dopamine: A framework for flexible Reinforcement Learning research"},"popularity":{"monthly_downloads":98516,"position":13077,"tier":"top_15000"},"security":{"n_vulnerabilities":0},"version":"4.1.2"}