File size: 2,338 Bytes
8c10cf2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
# MarkdownValidator Analysis Scripts

This directory contains Python scripts for analyzing the performance of various LLM models on the MarkdownValidator task.

## Scripts

### 1. analyze_retry_patterns.py
Analyzes RETRY patterns in execution paths, including:
- Error locations and frequencies
- RETRY counts and rates
- Error types by agent and node type

**Output files:**
- `retry_analysis.json` - Detailed retry statistics
- `retry_summary.csv` - Summary table
- `error_by_agent.csv` - Error breakdown by agent

### 2. evaluate_scores.py
Collects and analyzes score statistics from model executions.

**Output files:**
- `score_analysis.json` - Detailed score data
- `score_summary.csv` - Summary statistics (mean, min, max, perfect rate)

### 3. evaluate_success.py
Calculates success/failure rates for each model.

**Output files:**
- `success-finish_detailed_results.json` - Detailed execution results
- `success-finish_rate.csv` - Success rate summary

### 4. evaluate_trajectory.py
Evaluates execution trajectories against a reference trajectory using multiple metrics:
- Exact match
- In-order match
- Any-order match
- Precision
- Recall
- Single-tool use
- Path diversity (unique_path_ratio)
- Path entropy

**Output files:**
- `evaluation_results.csv` - Trajectory evaluation results

### 5. reference_trajectory.yaml
Configuration file defining the ideal reference trajectory for the MarkdownValidator task.

## Usage

Run each script from the MarkdownValidator directory:

```bash
# Analyze retry patterns
python3 analyze_retry_patterns.py

# Evaluate scores
python3 evaluate_scores.py

# Evaluate success rates
python3 evaluate_success.py

# Evaluate trajectories
python3 evaluate_trajectory.py --config reference_trajectory.yaml
```

## Models Evaluated

- GPT-5
- GPT-4o-mini
- DeepSeek-V3-1
- DeepSeek-R1
- Gemini-2.5-flash
- Gemini-2.5-flash-nothinking
- Qwen3-235b

## Requirements

- Python 3.7+
- pandas
- PyYAML

## Data Structure

Scripts expect data in the following structure:
```
/Users/wzr/TOSEM-2025/RESULTS/
├── {Model}/
│   └── MarkdownValidator/
│       └── test_results/
│           ├── session_001/
│           │   ├── execution_path.md
│           │   ├── execution_info.json
│           │   └── summary.json
│           └── ...
```