File size: 6,373 Bytes
b064a39
 
7fc4a6c
3b95ded
23e294b
 
 
 
 
 
 
 
 
 
 
 
 
b064a39
3b95ded
c2fd8ff
23e294b
 
c2fd8ff
7fc4a6c
3b95ded
23e294b
 
3b95ded
23e294b
b064a39
7fc4a6c
 
 
3b95ded
eaaacf5
3b95ded
 
 
 
 
7fc4a6c
b064a39
3b95ded
b064a39
da4759e
 
 
23e294b
0b77c6e
 
3b95ded
0b77c6e
23e294b
 
 
 
 
 
 
 
 
da4759e
23e294b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0b77c6e
23e294b
0b77c6e
b064a39
23e294b
 
 
 
0b77c6e
23e294b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eaaacf5
3b95ded
 
 
 
 
 
 
 
eaaacf5
3b95ded
 
 
 
eaaacf5
3b95ded
 
 
eaaacf5
3b95ded
 
eaaacf5
3b95ded
7fc4a6c
 
3b95ded
7fc4a6c
0b77c6e
3b95ded
7fc4a6c
da4759e
3b95ded
7fc4a6c
 
3b95ded
 
 
 
 
 
7fc4a6c
0b77c6e
da4759e
0b77c6e
 
 
3b95ded
23e294b
 
3b95ded
 
7fc4a6c
0b77c6e
23e294b
 
0b77c6e
 
b064a39
23e294b
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
import gradio as gr
import pandas as pd
from pathlib import Path
import logging
from datetime import datetime
import sys
import uuid
from typing import Dict, Any

# Add parent directory to path to import main
sys.path.append(str(Path(__file__).parent))
from main import (
    StorageManager, 
    EvaluationRequest,
    evaluate_model,
    PATHS
)

logging.basicConfig(level=logging.INFO)

# Initialize storage manager
storage_manager = StorageManager(PATHS)

def load_leaderboard_data():
    try:
        return pd.DataFrame(storage_manager.load('leaderboard'))
    except Exception as e:
        logging.error(f"Error loading leaderboard: {e}")
        return pd.DataFrame()

def format_leaderboard_df(df):
    if df.empty:
        return df
    
    display_df = pd.DataFrame({
        "Model": df["model"],
        "Average PER ⬇️": df["average_per"].apply(lambda x: f"{x:.4f}"),
        "Average PWED ⬇️": df["average_pwed"].apply(lambda x: f"{x:.4f}"),
        "GitHub": df["github_url"].apply(lambda x: f'<a href="{x}" target="_blank">Repository</a>' if x else "N/A"),
        "Submission Date": pd.to_datetime(df["submission_date"]).dt.strftime("%Y-%m-%d")
    })
    
    return display_df.sort_values("Average PER ⬇️")

def create_html_table(df):
    return df.to_html(escape=False, index=False, classes="styled-table")

def submit_evaluation(model_name: str, submission_name: str, github_url: str) -> str:
    if not model_name or not submission_name:
        return "⚠️ Please provide both model name and submission name."
    
    try:
        # Generate a task ID
        task_id = str(uuid.uuid4())
        
        # Create evaluation request
        request = EvaluationRequest(
            transcription_model=model_name,
            submission_name=submission_name,
            github_url=github_url if github_url else None,
            subset="test"
        )
        
        # Create task entry
        task = {
            "id": task_id,
            "model": model_name,
            "subset": "test",
            "submission_name": submission_name,
            "github_url": github_url,
            "status": "queued",
            "submitted_at": datetime.now().isoformat()
        }
        
        # Save task
        tasks = storage_manager.load('tasks')
        tasks.append(task)
        storage_manager.save('tasks', tasks)
        
        # Start evaluation in background
        import asyncio
        asyncio.run(evaluate_model(task_id, request))
        
        return f"βœ… Evaluation submitted successfully! Task ID: {task_id}"
    except Exception as e:
        return f"❌ Error: {str(e)}"

def check_status(query: str) -> Dict[str, Any]:
    if not query:
        return {"error": "Please enter a model name or task ID"}
    
    try:
        results = storage_manager.load('results')
        tasks = storage_manager.load('tasks')
        
        # First try to find by task ID
        result = next((r for r in results if r["task_id"] == query), None)
        task = next((t for t in tasks if t["id"] == query), None)
        
        # If not found, try to find by model name
        if not result:
            result = next((r for r in results if r["model"] == query), None)
        if not task:
            task = next((t for t in tasks if t["model"] == query), None)
        
        if result:
            # If we found results, return them
            return {
                "status": "completed",
                "model": result["model"],
                "subset": result["subset"],
                "num_files": result["num_files"],
                "average_per": result["average_per"],
                "average_pwed": result["average_pwed"],
                "detailed_results": result["detailed_results"],
                "timestamp": result["timestamp"]
            }
        elif task:
            # If we only found task status, return that
            return task
        else:
            return {"error": f"No results found for '{query}'"}
            
    except Exception as e:
        logging.error(f"Error checking status: {e}")
        return {"error": f"Error checking status: {str(e)}"}

with gr.Blocks(css="""
    .styled-table {
        width: 100%;
        border-collapse: collapse;
        margin: 25px 0;
        font-size: 0.9em;
        font-family: sans-serif;
        box-shadow: 0 0 20px rgba(0, 0, 0, 0.15);
    }
    .styled-table thead tr {
        background-color: #009879;
        color: #ffffff;
        text-align: left;
    }
    .styled-table th,
    .styled-table td {
        padding: 12px 15px;
    }
    .styled-table tbody tr {
        border-bottom: 1px solid #dddddd;
    }
""") as demo:
    gr.Markdown("# 🎯 Phonemic Transcription Model Evaluation Leaderboard")
    
    with gr.Tabs():
        with gr.TabItem("πŸ† Leaderboard"):
            leaderboard_html = gr.HTML(create_html_table(format_leaderboard_df(load_leaderboard_data())))
            refresh_btn = gr.Button("πŸ”„ Refresh")
            refresh_btn.click(
                lambda: gr.HTML.update(value=create_html_table(format_leaderboard_df(load_leaderboard_data()))),
                outputs=leaderboard_html
            )
        
        with gr.TabItem("πŸ“ Submit Model"):
            model_name = gr.Textbox(label="Model Name", placeholder="facebook/wav2vec2-lv-60-espeak-cv-ft")
            submission_name = gr.Textbox(label="Submission Name", placeholder="My Model v1.0")
            github_url = gr.Textbox(label="GitHub URL (optional)", placeholder="https://github.com/username/repo")
            submit_btn = gr.Button("Submit")
            result = gr.Textbox(label="Submission Status")
            
            submit_btn.click(
                fn=submit_evaluation,
                inputs=[model_name, submission_name, github_url],
                outputs=result
            )
        
        with gr.TabItem("πŸ“Š Model Status"):
            query = gr.Textbox(label="Model Name or Task ID", placeholder="Enter model name (e.g., facebook/wav2vec2-lv-60-espeak-cv-ft)")
            status_btn = gr.Button("Check Status")
            status_output = gr.JSON(label="Status")
            
            status_btn.click(
                fn=check_status,
                inputs=query,
                outputs=status_output
            )

if __name__ == "__main__":
    demo.launch()