ChatGPT Code Interpreterクローンを試す（Code Interpreter API編）

$ git clone https://github.com/shroominic/codeinterpreter-api

{
	"name": "Python 3",
	"image": "mcr.microsoft.com/devcontainers/python:1-3.11-bookworm",
	"features": {
		"ghcr.io/devcontainers-contrib/features/poetry:2": {}
	}
}

$ poetry install

$ cp .env.example .env

OPENAI_API_KEY=XXXXXXXXXX
VERBOSE=True

$ poetry run python examples/analyze_dataset.py 

> Entering new AgentExecutor chain...
⚙️ Running code: ```python
# First, let's load the dataset and take a look at it.
import pandas as pd

# Load the dataset
iris = pd.read_csv('iris.csv')

# Display the first few rows of the dataset
iris.head()
```

Invoking: `python` with `{'code': "# First, let's load the dataset and take a look at it.\nimport pandas as pd\n\n# Load the dataset\niris = pd.read_csv('iris.csv')\n\n# Display the first few rows of the dataset\niris.head()"}`


SepalLengthCm  SepalWidthCm  PetalLengthCm  PetalWidthCm      Species
0            5.1           3.5            1.4           0.2  Iris-setosa
1            4.9           3.0            1.4           0.2  Iris-setosa
2            4.7           3.2            1.3           0.2  Iris-setosa
3            4.6           3.1            1.5           0.2  Iris-setosa
4            5.0           3.6            1.4           0.2  Iris-setosa
⚙️ Running code: ```python
# Check the distribution of the different species in the dataset
iris['Species'].value_counts()
```

Invoking: `python` with `{'code': "# Check the distribution of the different species in the dataset\niris['Species'].value_counts()"}`
responded: {content}

Species
Iris-setosa        50
Iris-versicolor    50
Iris-virginica     50
Name: count, dtype: int64
⚙️ Running code: ```python
import seaborn as sns
import matplotlib.pyplot as plt

# Plot pairwise relationships in the dataset
sns.pairplot(iris, hue='Species')

# Show the plot
plt.show()
```

Invoking: `python` with `{'code': "import seaborn as sns\nimport matplotlib.pyplot as plt\n\n# Plot pairwise relationships in the dataset\nsns.pairplot(iris, hue='Species')\n\n# Show the plot\nplt.show()"}`
responded: {content}

Image image-d0b3f5a1-38a5-48a1-8719-72b2e27216f5.png got send to the user.The pairplot provides a detailed visual insight into the dataset. Here are some observations:

1. Iris-setosa seems to be separable from Iris-versicolor and Iris-virginica based on PetalLengthCm and PetalWidthCm.
2. Iris-versicolor and Iris-virginica are somewhat overlapping with each other.
3. SepalWidthCm and SepalLengthCm are not as useful for distinguishing between the species.

This kind of exploratory analysis is very useful when we want to build a predictive model later, as it helps us understand which features are most informative.

> Finished chain.
AI:  The pairplot provides a detailed visual insight into the dataset. Here are some observations:

1. Iris-setosa seems to be separable from Iris-versicolor and Iris-virginica based on PetalLengthCm and PetalWidthCm.
2. Iris-versicolor and Iris-virginica are somewhat overlapping with each other.
3. SepalWidthCm and SepalLengthCm are not as useful for distinguishing between the species.

This kind of exploratory analysis is very useful when we want to build a predictive model later, as it helps us understand which features are most informative.
<PIL.Image.Image image mode=RGB size=1137x986 at 0x7F72282AE810>

$ poetry run streamlit run frontend/app.py

> Entering new AgentExecutor chain...
⚙️ Running code: ```python
!pip install gtts
```

Invoking: `python` with `{'code': '!pip install gtts'}`


Collecting gtts
Downloading gTTS-2.3.2-py3-none-any.whl (28 kB)
Installing collected packages: gtts
Successfully installed gtts-2.3.2
⚙️ Running code: ```python
from gtts import gTTS
import os

text = "明日はいいお天気ですね。"
tts = gTTS(text, lang='ja')
tts.save("output.mp3")

os.listdir()
```

Invoking: `python` with `{'code': 'from gtts import gTTS\nimport os\n\ntext = "明日はいいお天気ですね。"\ntts = gTTS(text, lang=\'ja\')\ntts.save("output.mp3")\n\nos.listdir()'}`


['output.mp3', 'iris.csv']
ファイル「output.mp3」が生成されました。

> Finished chain.

 $ poetry run pip freeze | grep -i gtts
gTTS==2.3.2

INFO: Using a LocalBox which is not isolated.
      This is only for testing and development.
      Make sure to put an API-Key in production.