OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Developer Tools & SDKs

Unirl — Tencent-Hunyuan

UniRL is an open-source Python framework for reinforcement-learning post-training of unified multimodal models, with training entrypoints, model recipes, algorithms, rollout engines and distributed-runtime components.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Unirl — Tencent-Hunyuan — OpenViST project profile in AI Developer Tools & SDKs.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use UniRL when developing or adapting multimodal model training workflows that need a configurable RL post-training loop, or when exploring the repository’s documented model and algorithm recipes.

Who it is for

  • AI developers building reinforcement-learning training workflows for multimodal models
  • Research teams experimenting with model-specific recipes and reinforcement-learning algorithms

Documented uses

  • Coordinate sample generation, scoring, advantage calculation and policy updates in a multimodal RL post-training loop
  • Use documented model recipes for image and video generation or editing tasks
  • Configure domain-specific trainers and connect them to rollout engines, algorithms and reward services
Before you choose

Requirements & limits

The public GitHub repository is identified as open source, and its README displays an Apache-2.0 license badge. The README specifies Python 3.12 or later.

  • Python 3.12 or later is specified in the repository README.
  • The framework connects training entrypoints and Hydra example configurations to domain trainers and pluggable components.
  • The repository documents a single-node SD3 diffusion example and refers users to separate multi-node launch guidance.
  • Model and algorithm support are described as independent dimensions; support should be checked against the repository’s model matrix and individual recipes.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Check and launch the single-node SD3 example

Resolve the diffusion example configuration, then launch its single-node experiment.

Before you start

  • Python 3.12 or later
  • Dependencies installed according to INSTALL.md
  • The UniRL repository and its example files

Steps

  1. Follow INSTALL.md to install dependencies.

  2. Run the configuration check: python -m unirl.train_diffusion --config-name=diffusion/sd3/sd3_trainside --cfg job --resolve.

  3. Launch the example with bash examples/run_experiment_single_node.sh diffusion/sd3/sd3_trainside.

Expected outcome

The documented single-node Stable Diffusion 3.5 training example is launched after its configuration is resolved.

Guide limitations

  • The README points to the installation guide for dependency details.
  • This recipe is a single-node diffusion example; the repository points to separate guidance for multi-node launches.
  • The listed steps describe the project’s instructions, not an independently verified run.
Documentation 1 ↗Documentation 2 ↗
Evidence and transparency

Sources & review

Official repository ↗Guide documentation ↗
Source record details
Source availability
Active
Sources checked
18 October 2026
Original website title
Tencent-Hunyuan/UniRL · Official homepage title · retained in its source language
How we review sourcesSuggest a correction