File size: 3,321 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 | # SQLAssistant-A2A Evaluation Scripts
This directory contains evaluation scripts for analyzing the SQLAssistant-A2A project results across different language models.
## Scripts
### 1. analyze_retry_patterns.py
Analyzes RETRY patterns in execution paths, including:
- **Orchestrator RETRY**: Framework-level retries (e.g., `crew_execution (retry N)`)
- **Business RETRY**: Application-level retries (e.g., `generate_sql (business_retry N)`)
**Output Files:**
- `retry_analysis.json` - Detailed retry statistics
- `retry_summary.csv` - Summary statistics by model
- `error_by_agent.csv` - Error location statistics by agent
- `business_retry_chapters.csv` - Business retry location statistics
**Usage:**
```bash
python3 analyze_retry_patterns.py
```
### 2. evaluate_scores.py
Collects and analyzes score statistics from execution logs.
**Output Files:**
- `score_analysis.json` - Detailed score data
- `score_summary.csv` - Summary statistics by model
**Usage:**
```bash
python3 evaluate_scores.py
```
### 3. evaluate_success.py
Calculates success rates based on whether required tools were executed.
**Success Criteria:** Execution of `get_database_schema` tool (required to understand database structure)
**Output Files:**
- `success_detailed_results.json` - Detailed results by session
- `success_rate.csv` - Success rate summary by model
**Usage:**
```bash
python3 evaluate_success.py
```
### 4. evaluate_trajectory-Filter_Tools.py
Evaluates execution trajectories using 6 metrics:
1. **Exact match** - Complete trajectory match
2. **In-order match** - Required steps appear in order
3. **Any-order match** - All required steps present
4. **Precision** - Correctness of predicted steps
5. **Recall** - Coverage of required steps
6. **Single-tool use** - Individual tool usage rates
**Configuration:**
- Uses `reference_trajectory.yaml` for ground truth trajectory definition
**Output Files:**
- `evaluation_results.csv` - Evaluation metrics by model
- `evaluation_results.md` - Formatted results report
- `any_order_match_failures.csv` - Failure analysis (with `--diagnose-failures`)
**Usage:**
```bash
python3 evaluate_trajectory-Filter_Tools.py --config reference_trajectory.yaml
python3 evaluate_trajectory-Filter_Tools.py --diagnose-failures # With failure analysis
```
## Configuration
### reference_trajectory.yaml
Defines the ground truth execution trajectory for the SQLAssistant-A2A project, including:
- Reference trajectory steps
- Target tools for evaluation
- Model list
- Extract types (SPAN, Chain, Agent, LLM, Tool)
## Models Evaluated
- DeepSeek-R1
- DeepSeek-V3-1
- GPT-4o-mini
- GPT-5
- Gemini-2.5-flash
- Gemini-2.5-flash-nothinking
- Qwen3-235b
## Data Structure
Scripts expect the following directory structure:
```
/Users/wzr/TOSEM-2025/RESULTS/
├── <model_name>/
│ └── SQLAssistant-A2A/
│ └── test_results/
│ ├── session_1/
│ │ ├── execution_path.md
│ │ └── execution_log.json
│ ├── session_2/
│ └── ...
```
## Requirements
- Python 3.x
- pandas
- PyYAML
## Notes
- All scripts automatically process all configured models
- Results are saved in the same directory as the scripts
- Scripts handle missing data gracefully and report warnings
|