File size: 3,868 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
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
# SQLAssistant-H_A2A Evaluation Scripts

This directory contains evaluation scripts for the SQLAssistant-H_A2A project, which analyzes the execution trajectories and performance metrics of different AI models.

## Files

### Configuration
- `reference_trajectory.yaml` - Reference trajectory configuration defining the expected execution path for the SQLAssistant workflow

### Evaluation Scripts
- `evaluate_trajectory-Filter_Tools-mix.py` - Main trajectory evaluation script that computes 6 trajectory metrics
- `evaluate_success.py` - Success rate analysis based on required tool execution
- `evaluate_scores.py` - Score statistics collection from execution logs
- `analyze_retry_patterns.py` - RETRY pattern analysis including business retries and orchestrator retries

## Trajectory Metrics

The evaluation framework computes the following 6 metrics:

1. **Exact Match** - Predicted trajectory must exactly match the reference trajectory
2. **In-Order Match** - Reference trajectory must be a subsequence of predicted trajectory
3. **Any-Order Match** - Predicted trajectory must contain all necessary actions (order-agnostic)
4. **Precision** - Ratio of correct actions in predicted trajectory
5. **Recall** - Ratio of reference actions covered by predicted trajectory
6. **Single-Tool Use** - Usage rate of specific tools

Additional diversity metrics:
- **Unique Path Ratio** - Number of unique trajectories / total samples
- **Path Entropy** - Shannon entropy of trajectory distribution

## Usage

### Trajectory Evaluation
```bash
python3 evaluate_trajectory-Filter_Tools-mix.py --config reference_trajectory.yaml
```

Options:
- `--config` - Path to reference trajectory configuration file (YAML format)
- `--base-dir` - RESULTS directory path (defaults to two levels up from script location)
- `--output` - Output CSV file path
- `--diagnose-failures` - Diagnose any_order_match failure reasons and generate report

### Success Rate Analysis
```bash
python3 evaluate_success.py
```

Success criterion: Whether the `get_database_schema` tool was executed (required for understanding database structure).

### Score Statistics
```bash
python3 evaluate_scores.py
```

Collects and analyzes scores from execution logs.

### Retry Pattern Analysis
```bash
python3 analyze_retry_patterns.py
```

Analyzes two types of retries:
1. **Orchestrator RETRY** - Framework-level retries (e.g., `[SPAN] crew_execution (retry N)`)
2. **BUSINESS-RETRY** - Business logic retries (e.g., `[SPAN] generate_sql (business_retry N)`)

## Models Evaluated

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

## Project Structure

The SQLAssistant-H_A2A project uses a hybrid architecture combining:
- **CrewAI** for SQL generation stage
- **LangGraph** for compliance checking stage
- **AutoGen** for result interpretation stage

### Workflow Stages

1. **Generate SQL** - CrewAI-based SQL generation with two agents:
   - Expert SQL Query Generator
   - Senior SQL Code Reviewer

2. **Check Compliance** - LangGraph orchestration for compliance validation

3. **Interpret Results** - AutoGen-based result interpretation and SQL execution

## Output Files

- `evaluation_results.csv` - Trajectory evaluation results for all models
- `success_rate.csv` - Success rate summary
- `success_detailed_results.json` - Detailed success analysis
- `score_summary.csv` - Score statistics summary
- `score_analysis.json` - Detailed score analysis
- `retry_summary.csv` - Retry pattern summary
- `retry_analysis.json` - Detailed retry analysis
- `error_by_agent.csv` - Error statistics by agent
- `business_retry_chapters.csv` - Business retry location statistics
- `any_order_match_failures.csv` - Failure diagnosis (if enabled)

## Requirements

- Python 3.x
- pandas
- PyYAML
- pathlib
- collections
- json
- csv
- re
- math