Vighnesh commited on
Commit
d771897
ยท
1 Parent(s): 95dc191

Fix: add gradio to pyproject.toml deps, update README structure to match actual files

Browse files
Files changed (2) hide show
  1. README.md +24 -14
  2. pyproject.toml +1 -0
README.md CHANGED
@@ -77,30 +77,38 @@ Rewards provide partial progress signals throughout the trajectory:
77
 
78
  ```
79
  support_ticket_env/
80
- โ”œโ”€โ”€ __init__.py # Package exports
81
- โ”œโ”€โ”€ models.py # SupportAction, SupportObservation, SupportState
82
- โ”œโ”€โ”€ tickets.py # Ticket dataset with ground-truth labels
83
- โ”œโ”€โ”€ graders.py # Reward/grader functions for all 3 tasks
84
- โ”œโ”€โ”€ client.py # EnvClient subclass
85
- โ”œโ”€โ”€ baseline.py # Baseline inference script
86
- โ”œโ”€โ”€ openenv.yaml # Environment metadata
87
- โ”œโ”€โ”€ Dockerfile # Container definition
 
 
 
 
 
 
 
 
88
  โ””โ”€โ”€ server/
89
- โ”œโ”€โ”€ app.py # FastAPI entry point
90
- โ””โ”€โ”€ support_environment.py # Environment logic
 
91
  ```
92
 
93
  ## Setup
94
 
95
  ```bash
96
  # Install dependencies
97
- pip install openenv-core fastapi uvicorn pydantic gradio openai
98
 
99
  # Run locally
100
- cd support_ticket_env
101
- uvicorn server.app:app --host 0.0.0.0 --port 7860
102
 
103
- # Docker
104
  docker build -t support-ticket-env .
105
  docker run -p 7860:7860 support-ticket-env
106
 
@@ -108,6 +116,8 @@ docker run -p 7860:7860 support-ticket-env
108
  python run_tests.py
109
  ```
110
 
 
 
111
  ## Baseline Scores
112
 
113
  Measured with `gpt-4o-mini`, seeds `[42, 7, 123]`:
 
77
 
78
  ```
79
  support_ticket_env/
80
+ โ”œโ”€โ”€ __init__.py # Package exports
81
+ โ”œโ”€โ”€ models.py # SupportAction, SupportObservation, SupportState
82
+ โ”œโ”€โ”€ tickets.py # Ticket dataset with ground-truth labels
83
+ โ”œโ”€โ”€ graders.py # Reward/grader functions for all 3 tasks
84
+ โ”œโ”€โ”€ client.py # EnvClient subclass
85
+ โ”œโ”€โ”€ baseline.py # Baseline inference script
86
+ โ”œโ”€โ”€ get_baseline.py # Fetch & save baseline results
87
+ โ”œโ”€โ”€ gradio_ui.py # Interactive Gradio playground UI
88
+ โ”œโ”€โ”€ make_chart.py # Plot training reward curves
89
+ โ”œโ”€โ”€ plot_results.py # Visualise evaluation results
90
+ โ”œโ”€โ”€ grpo_results.png # GRPO training results chart
91
+ โ”œโ”€โ”€ reward_chart.png # Reward curve chart
92
+ โ”œโ”€โ”€ openenv.yaml # Environment metadata
93
+ โ”œโ”€โ”€ Dockerfile # Container definition
94
+ โ”œโ”€โ”€ train_sft.ipynb # Step 1: SFT pre-training notebook
95
+ โ”œโ”€โ”€ train_grpo.ipynb # Step 2: GRPO fine-tuning notebook
96
  โ””โ”€โ”€ server/
97
+ โ”œโ”€โ”€ app.py # FastAPI entry point (+ Gradio UI mounted at /playground)
98
+ โ”œโ”€โ”€ support_environment.py # Environment logic
99
+ โ””โ”€โ”€ requirements.txt # Server dependencies
100
  ```
101
 
102
  ## Setup
103
 
104
  ```bash
105
  # Install dependencies
106
+ pip install openenv-core fastapi uvicorn pydantic gradio openai pyyaml
107
 
108
  # Run locally
109
+ uvicorn support_ticket_env.server.app:app --host 0.0.0.0 --port 7860
 
110
 
111
+ # Or via Docker
112
  docker build -t support-ticket-env .
113
  docker run -p 7860:7860 support-ticket-env
114
 
 
116
  python run_tests.py
117
  ```
118
 
119
+ > ๐ŸŽฎ **Playground UI** available at `http://localhost:7860/playground` once the server is running.
120
+
121
  ## Baseline Scores
122
 
123
  Measured with `gpt-4o-mini`, seeds `[42, 7, 123]`:
pyproject.toml CHANGED
@@ -16,6 +16,7 @@ dependencies = [
16
  "pydantic>=2.0.0",
17
  "openai>=1.0.0",
18
  "pyyaml>=6.0",
 
19
  ]
20
 
21
  [project.optional-dependencies]
 
16
  "pydantic>=2.0.0",
17
  "openai>=1.0.0",
18
  "pyyaml>=6.0",
19
+ "gradio>=4.0.0",
20
  ]
21
 
22
  [project.optional-dependencies]