Skip to content

Commit db00fc4

Browse files
committed
Update sela README
1 parent 3a47b4e commit db00fc4

File tree

1 file changed

+35
-36
lines changed

1 file changed

+35
-36
lines changed

metagpt/ext/sela/README.md

Lines changed: 35 additions & 36 deletions
Original file line numberDiff line numberDiff line change
@@ -19,11 +19,6 @@ You can either download the datasets from the link or prepare the datasets from
1919

2020
## 2. Configurations
2121

22-
### Data Config
23-
24-
- **`datasets.yaml`:** Provide base prompts, metrics, and target columns for respective datasets.
25-
- **`data.yaml`:** Modify `datasets_dir` to the base directory of all prepared datasets.
26-
2722
### LLM Config
2823

2924
```yaml
@@ -34,13 +29,7 @@ llm:
3429
api_key: sk-xxx
3530
temperature: 0.5
3631
```
37-
38-
39-
## 3. SELA
40-
41-
### Run SELA
42-
43-
#### Setup
32+
### Setup
4433

4534
```bash
4635
pip install -e .
@@ -50,33 +39,43 @@ cd metagpt/ext/sela
5039
pip install -r requirements.txt
5140
```
5241

53-
#### Quick Start
42+
## 3. Quick Start
5443

55-
- **Example : Running SELA on the House Price Prediction Task**
56-
- To run the project, simply execute the following command:
57-
```bash
58-
python run_sela.py
59-
```
60-
- Explanation of `run_sela.py`:
61-
```bash
62-
requirement = ("Optimize dataset using MCTS with 10 rollouts. "
63-
"This is a 05_house-prices-advanced-regression-techniques dataset."
64-
"Your goal is to predict the target column `SalePrice`."
65-
"Perform data analysis, data preprocessing, feature engineering, and modeling to predict the target."
66-
"Report rmse on the eval data. Do not plot or make any visualizations.")
67-
data_dir = "Path/to/dataset"
68-
69-
sela = SELA()
70-
await sela.run(requirement, data_dir)
71-
```
72-
73-
#### Running Experiments
44+
### Example : Running SELA on the House Price Prediction Task
7445

75-
- **Examples:**
46+
- **To run the project, simply execute the following command**
7647
```bash
48+
python run_sela.py
49+
```
50+
51+
- **Explanation of `run_sela.py`**
52+
```bash
53+
requirement = ('''
54+
Optimize dataset using MCTS with 10 rollouts.
55+
This is a 05_house-prices-advanced-regression-techniques dataset.
56+
Your goal is to predict the target column `SalePrice`.
57+
Perform data analysis, data preprocessing, feature engineering, and modeling to predict the target.
58+
Report rmse on the eval data. Do not plot or make any visualizations.''')
59+
data_dir = "Path/to/dataset"
60+
61+
sela = SELA()
62+
await sela.run(requirement, data_dir)
63+
```
64+
65+
## 4. SELA Reproduction Details
66+
67+
### Data Config
68+
- **`datasets.yaml`:** Provide base prompts, metrics, and target columns for respective datasets.
69+
- **`data.yaml`:** Modify `datasets_dir` to the base directory of all prepared datasets.
70+
71+
### Run SELA
72+
73+
#### Examples
74+
75+
```bash
7776
python run_experiment.py --exp_mode mcts --task titanic --rollouts 10
7877
python run_experiment.py --exp_mode mcts --task house-prices --rollouts 10 --low_is_better
79-
```
78+
```
8079

8180
#### Parameters
8281

@@ -98,7 +97,7 @@ pip install -r requirements.txt
9897

9998
### Ablation Study
10099

101-
**RandomSearch**
100+
#### RandomSearch
102101

103102
- **Use a single insight:**
104103
```bash
@@ -110,7 +109,7 @@ pip install -r requirements.txt
110109
python run_experiment.py --exp_mode rs --task titanic --rs_mode set
111110
```
112111

113-
## 4. Citation
112+
## 5. Citation
114113
Please cite our paper if you use SELA or find it cool or useful!
115114

116115
```bibtex

0 commit comments

Comments
 (0)