ChatGPT exploration: CodeInterpreter deep usage, advanced guide

Tech 2023-07-12 07:57:49 Source: Network

Code Interpreter Code Interpreter BetaChatGPT ChatGPT Code Interpreter&Data analysisBuilt in Python programming capability: This model has a built-in Python interpreter that can execute Python code.Support file upload and download: able to upload files to the workspace of the current session and download work results

Code Interpreter

Code Interpreter BetaChatGPT ChatGPT Code Interpreter&Data analysis

Built in Python programming capability: This model has a built-in Python interpreter that can execute Python code.
Support file upload and download: able to upload files to the workspace of the current session and download work results.
Isolated execution environment: The interpreter plugin works in a secure operating environment, including a firewall and temporary disk space.
Persistent session: The code is evaluated by the interpreter plugin in a continuous session, where the session remains active during the chat session and can be built through multiple calls (destroyed upon timeout).
Code Interpreter Data analysisvisualization
Encourage beginners to actively try and discover more interesting features.

session time out
session time out
This codeinsider (beta) chatlastimeout. Youmaycontinue the conversation, but previous files, links, and codeblocksbelowmaynotworkasexpected
Code Interpreter
CodeInterpretersession expired
Code Interpreter

visualization

Code Interpreter PythonPythonMatplotlibqrocdeGenerate QR codePILGraphvizflow chart

In the field of data analysis and processing, Matplotlib is undoubtedly a heavyweight player. It is a Python 2D drawing library that can generate high-quality images of various publication quality levels. It supports multiple platforms and provides many Graphics library tools. Here are some commonly used Matplotlib graphics and their application scenarios:

Line Plot: A line plot is an ideal choice to display the trend of data over time or continuous values, such as stock price trends, climate change, etc.

Bar Chart: When comparing data from multiple categories, a bar chart is an excellent choice. For example, comparing the population of different cities.

HorizontalBarChart: Similar to a bar chart, but in a horizontal direction. Suitable for situations where category labels are too long or there are too many categories.

Stacked Bar Chart: A stacked bar chart can display the total amount of each category and the proportion of its subcategories, such as the company's quarterly total sales and the proportion of sales of each product.

PieChart: A pie chart is often used to represent the proportion of each component in the overall population, such as the proportion of a company's revenue sources.

Histogram: A histogram is suitable for displaying the distribution of a set of data, such as the distribution of student grades.

Box plot: Box plot (also known as Box Plot) is used to show the five digit summary of a group of data, namely the minimum value, the lower Quartile, the median, the upper Quartile and the maximum value.

Violin Plot: Violin Plot is a combination of Box plot and Kernel density estimation plot, which can show both the distribution of data and the probability density of data.

ContourPlot: ContourPlot is usually used to display 3D data. Two dimensions are represented on a plane, and the third dimension is represented by Contour line.

Heatmap: Heatmap is often used to represent the correlation or distribution of data, such as the heat map of a webpage click.

BubbleChart: Bubble chart is an extension of Scatter plot, where the size and color of each point can be used to represent more dimensional data.

Radar Chart: Radar charts are often used to display multivariate data, especially suitable for comparing the performance of different objects on various characteristics.

Three Dimensional Plot: A three-dimensional plot is used to display the relationships between three numerical variables, such as the relationships between x, y, and z dimensions in three-dimensional space.

Area Plot: Area plots are very suitable for representing quantities that change over time and can compare differences between different categories.

Error Bar Plot: The error bar plot is used to show the uncertainty of data, such as experimental data represented by standard deviation or Standard error.

Fill Between Plot: A fill between plot can clearly represent the differences between two data series or the uncertainty of the same series of data.

Step Plot: A step plot is very suitable for representing discrete, discontinuous events, or threshold changes.

Composite Plot: A composite plot is a combination of multiple graphs, such as displaying both line and column charts in the same chart. It is suitable for scenarios where multiple related indicators need to be displayed simultaneously.

Candlestick Chart: Candlestick charts are often used in financial analysis to represent the opening, closing, highest, and lowest prices of stocks, options, etc.

WordCloud: WordCloud is commonly used in text data to display the most frequently occurring words in the text.

Polar Plot: Polar plot is used to represent data in Polar coordinate system, such as radar plot, wind rose plot, etc.

You should choose these graphics based on the characteristics of the data and the information you want to convey. In practical applications, it may be necessary to adjust the parameters of the chart according to specific needs to better display and interpret the data. If you want to learn more, you can refer to the official documentation of Matplotlib.

Use Cases

visualization Kaggle Kaggle Data analysissummary

Summarize and suggest

Prompt: Please analyze the prompts. csv file and display the analysis results through appropriate charts. Meanwhile, based on the analysis results, provide a high-quality guide for writing a prompt. You can use word clouds for high-frequency word statistics

Please download and upload the prompts. csv file from f/awesome chat prompt, and then ask for the above prompts one by one. Based on these operations, we can determine the best strategy for writing a prompt:

Clarify your needs: Use words such as want, need, help, tell, etc. to clearly indicate what you want ChatGPT to do.
Role setting: Using expressions such as "actasa" can help ChatGPT determine the role you want it to play.
Simplicity and clarity: Try to make your prompts as concise and clear as possible, and most prompts should be limited to less than 200 characters in length.
Indicate specific actions: If possible, use clear verbs such as use, create, design, play, etc. to express the specific actions you expect ChatGPT to perform.

video processing

Prompt
Convert it to MP4 format and play it (uploaded video file: test. mov)
Display the total duration of the video and capture an image from the 1st/10th second of the video to display it
Show this picture

test.mov .mp4 .gif Code Interpreter

Image processing

Promot displays the size of the original image, cuts the largest square from the center, and displays the cropped image. Use a size of 576x576 as the sliding window of the original image, moving 1/50 of the image width each time, and make the sliding process into a loop playing gif. Convert the image to black and white and display it. Extract colors from the original image to generate a palette, and display color values and palette images. Convert the original image to JPG format and provide a download link, while displaying the original image

After uploading the image, follow the prompts above in sequence. You can not only obtain information about the image, but also perform editing operations on the image (such as cropping, generating gifs, adjusting colors, converting formats, etc.). It should be noted that if the session expires, you will need to upload the image again, and there may be memory limitations when converting or displaying the image.

Original Image (StableDiffusion)

Grayscale processing

CodeInterpreter uses sliding window mode to output images as GIF animations

PDF processing

Prompt to upload the file, and then enter the above prompts in sequence. You can obtain an overview of the main content of the file, analyze the entire text, and try to extract some images as supplementary explanations. In addition, you can also merge the first three pages of the PDF file into one image for display

Upload PDF files and follow the prompts one by one. GPT can help you analyze the text content in a file, but attempting to extract some images may fail due to high complexity. However, if you need to merge multiple PDF pages into one image, this feature performs quite well.

Data analysis

Promptvisualization

countries-table.csv visualization

Generate QR code

Prompt, please provide the following website address https://github.com/lencx Convert to QR code and display it, and modify it using the following code:
Fromqrcode. image. styles. colortaskimportRadialGradiantColorMask
Qr. make_ Image (image_factory=StyledPilImage, color_mask=RadialGradientColorMask())

promptGenerate QR code qrcode GPT

flow chart

Prompt: Enter the above prompts one by one, and you can see how GPT attempts to generate innovative and complex visual arts. However, you may find that existing attempts may not meet your expectations - they may lack the artistic beauty you expect, or they may not be entirely innovative. You can continue to provide guidance, such as "exploring 3D and perspective effects" or "finding artistic aesthetics similar to fractal", to help the model better meet your needs.
From PILimportImage
Img=Image. open ('/mnt/data/browser_work_process. png')
Img. show()

flow chart

How Browser Works

Visual creativity

Following the instructions above, you can try to explore the innovative, complex, and mathematical art tasks that CodeInterpreter can accomplish as much as possible. However, please note that these tasks need to be innovative and brand new, rather than just variations on existing works. For example, ChaosGame, Cellular automation, and Fractal are all innovative projects, but they already exist. In order to seek innovation, we may need to explore new areas such as 3D and perspective effects. If the current direction cannot achieve the goal, it is advisable to change your mindset and continue exploring. Please note that we are looking for works with artistic aesthetics, such as those like fractals.

Code Interpreter visualization

A Tree with Gradual Fractals

summary

Code Interpreter
Code Interpreter Python
Code Interpreter ChatGPT GPT-4 Data analysis
Code Interpreter
In addition, the temporary memory space allocated by the Python interpreter is limited. If you try to perform some complex and time-consuming tasks, you may fail due to the task being too large. It is more suitable for data processing and analysis, and may encounter difficulties in file format conversion tasks that may run out of memory due to large files.

Disclaimer: The content of this article is sourced from the internet. The copyright of the text, images, and other materials belongs to the original author. The platform reprints the materials for the purpose of conveying more information. The content of the article is for reference and learning only, and should not be used for commercial purposes. If it infringes on your legitimate rights and interests, please contact us promptly and we will handle it as soon as possible! We respect copyright and are committed to protecting it. Thank you for sharing.(Email:[email protected])

Mobile advertising space rental

Tag: ChatGPT exploration CodeInterpreter deep usage advanced guide