Spaces:
Sleeping
Sleeping
Commit ยท
5567f49
1
Parent(s): 35f712a
feat(ui): add comprehensive Quick Start guide to Gradio dashboard
Browse files
app.py
CHANGED
|
@@ -539,6 +539,39 @@ with gr.Blocks(title="Codebase Navigation & Repair โ OpenEnv v4") as demo:
|
|
| 539 |
|
| 540 |
with gr.Tabs():
|
| 541 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 542 |
# โโ Tab 1: Interactive โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
|
| 543 |
with gr.TabItem("๐ฎ Interactive"):
|
| 544 |
with gr.Row():
|
|
|
|
| 539 |
|
| 540 |
with gr.Tabs():
|
| 541 |
|
| 542 |
+
# โโ Tab 0: Quick Start Guide โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
|
| 543 |
+
with gr.TabItem("๐ Quick Start Guide"):
|
| 544 |
+
gr.Markdown("""
|
| 545 |
+
### Welcome to Codebase Navigation & Repair โ OpenEnv v4
|
| 546 |
+
|
| 547 |
+
This interactive dashboard allows you to experience the environment infrastructure, run simulations, and analyze advanced agent logic.
|
| 548 |
+
|
| 549 |
+
#### ๐ Step-by-Step Evaluation Guide:
|
| 550 |
+
|
| 551 |
+
1. **Initialize an Episode**
|
| 552 |
+
- Navigate to the **๐ค Run Agent** tab.
|
| 553 |
+
- Select a task (`task1`, `task2`, or `task3`) and click **"Run Agent"**.
|
| 554 |
+
- *This simulates an AI executing an episode dynamically against the environment and stores the trajectory.*
|
| 555 |
+
|
| 556 |
+
2. **Trigger Advanced Intelligence Diagnostics (v3/v4 Features)**
|
| 557 |
+
- Go to **๐งช Causal Probe** and click it to evaluate if the agent truly understood the bug, or if it was just pattern-matching.
|
| 558 |
+
- Go to **๐ญ Counterfactual** to run mutation tests and analyze the brittleness of the agent's logic.
|
| 559 |
+
- Go to **๐ Confidence** to see if the agent over-explored or submitted too early.
|
| 560 |
+
- Go to **๐ง Intelligence** to execute failure classification and strategy detection.
|
| 561 |
+
|
| 562 |
+
3. **Visualize the Thought Process**
|
| 563 |
+
- Head over to the **๐ 3D Visualizer** tab.
|
| 564 |
+
- Click **"Load / Refresh Visualizer"**.
|
| 565 |
+
- Using Three.js, this generates a dynamic 3D web of exactly how the agent traversed the repository files (cubes) and tests (prisms).
|
| 566 |
+
|
| 567 |
+
4. **Experiment Manually**
|
| 568 |
+
- Want to play the game yourself? Go to the **๐ฎ Interactive** tab.
|
| 569 |
+
- Click **Reset Environment**, then use the dropdowns to `read_file`, `write_file`, and finally `submit` to grade yourself.
|
| 570 |
+
|
| 571 |
+
5. **REST API / CLI Runner**
|
| 572 |
+
- The entire platform operates out of incredibly fast, natively compliant REST endpoints. Check the **๐ API** tab for standard cURL routing.
|
| 573 |
+
""")
|
| 574 |
+
|
| 575 |
# โโ Tab 1: Interactive โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
|
| 576 |
with gr.TabItem("๐ฎ Interactive"):
|
| 577 |
with gr.Row():
|