Skip to content

[Tunix] Replace HTML-based image centering with standard g3doc Markdown image syntax. #1503

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
copybara-service wants to merge 1 commit into
base: main
Choose a base branch
from
+4 −12
Open

[Tunix] Replace HTML-based image centering with standard g3doc Markdown image syntax. #1503

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 3 additions & 9 deletions docs/agentic_rl.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,9 +4,7 @@

## Architecture

<p align="center">
<img src="images/agentic_rollout_pipeline.png" alt="Trajectory Collect Engine Overview" width="100%">
</p>
![Trajectory Collect Engine Overview](images/agentic_rollout_pipeline.png)

## Core Components

Expand Down Expand Up @@ -65,9 +63,7 @@ calls in parallel for efficiency.

### Agent/Environment interaction

<p align="center">
<img src="images/agentic_agent:env.png" alt="Batch vs Async Rollout" width="80%">
</p>
![Batch vs Async Rollout](images/agentic_agent:env.png)

--------------------------------------------------------------------------------

Expand Down Expand Up @@ -116,9 +112,7 @@ lock ensures that rollouts (`acquire_rollout`) are temporarily paused when a
weight sync (`acquire_weight_sync`) is requested, preventing agents from
generating trajectories with stale parameters.

<p align="center">
<img src="images/batch_vs_async_rollout.png" alt="Batch vs Async Rollout" width="50%">
</p>
![Batch vs Async Rollout](images/batch_vs_async_rollout.png)

### Trajectory Batching and Grouping

Expand Down
4 changes: 1 addition & 3 deletions docs/design.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -123,9 +123,7 @@ training agents that can perform multi-turn reasoning and interact with external
tools. The design follows a standard RL paradigm where an **Agent** interacts
with an **Environment** over multiple steps to complete a task.

<p align="center">
<img src="images/agentic_rollout_pipeline.png" alt="Agentic RL Flow" width="80%">
</p>
![Agentic RL Flow](images/agentic_rollout_pipeline.png)

The core design supports agents that engage in **multi-turn conversations**,
breaking down complex problems into sequential steps of reasoning, tool
Expand Down
Loading