I just got a suggestion to use ace_tools as well from GPT-4o, so the issue remains. It was suggesting I use it display the dataframe, but I could easily just display it directly
1 Like
This is still an issue. I also just got several recommendations to use it. I was using custom GPTs, Which I assume means they’re now powered by GPT-4o… maybe?
1 Like
ace_tools got me here.
The generated code is also in the context of Pandas.
Display the final transcribed segments
import ace_tools as tools; tools.display_dataframe_to_user(… - it looked a bit weird so I Googled “ace_tools”
This is still happening. Here’s my prompt:
I want to compare two Huggingface datasets, and see a table which shows:
- common fields
- fields only in left
- fields only in right
Here are the two datasets:
ds_opyate["train"]
ds_ibm["train"]
And at the end of the completion, the ace_tools recommendation similar to OP.
1 Like
This is still happening.
My prompt was for web scraping code for publicly available information about companies as a test for 4o, but the suggested code ended with:
Display DataFrame to user
import ace_tools as tools; tools.display_dataframe_to_user(name=“Companies”, dataframe=df)
PyPI now contains two ace-tools links.
ace-tools 0.0 is an empty place holder but does have the use of
pip install ace-tools
this is described above by goksu.ismal98
A second library also appeared on the 12 July 2024
ace-tools-open 0.1.0
Open implementation of ace_tools referenced in GPT4o.
pip install ace_tools_open
There is currently no risk with the code at 29/7/2024
1 Like
I just had the same problem with gpt-4o
Nope! It happened to with ChatGPT 4 too.
same exact thing just happened to me, I was trying to visualize data that I was scraping and saved as json. There were so dictionaries of dictionaries, and I was having a really hard time visualizing it, but at the same time had to present it to my boss. Anyways, I asked chatGPT and the code that it ended up using had this line at the bottom of the script: import ace_tools as tools; tools.display_dataframe_to_user(name=“Inspection Data”, dataframe=df). Never saw this before yesterday and I have been doing a lot of web scraping over the past year and never came across it anywhere else.
Thank you for the information.
1 Like
Just happened to me as well
This just happened to me too. You all saved me here with this form. I googled the issue and there were more results than just here. I am really happy this was here and I did not download something. I like to think I would not download a random install but the fact that I was looking is concerning. Thank you for writing here!! This is what originally prompted me: “It appears that the ace_tools module is not available in your local environment, which makes sense as it’s a custom tool for this platform.” When I asked more question GPT told me I did not need Ace Tools and wrote additional code without it. My search for Ace Tools was mostly out of curiosity.
When I asked where to get Ace Tools GPT gave me an answer that was not going on the internet to look for an install.
GPT
"ace_tools is a custom module used within this platform for displaying data frames and other outputs in a user-friendly manner. It is not a publicly available Python package that you can install via pip. Instead, it is part of the internal tools used here to help visualize data directly within our interactions.
For your local environment, you can achieve similar functionality using standard Python libraries. Here’s how you can modify the script to save and display the data without ace_tools:"
1 Like
This is very helpful information, thank you very much!
1 Like
中国大学生在2024的数学建模比赛中也遇到了
添加决策结果到DataFrame
df[‘最优决策’] = decisions
使用工具显示结果
import ace_tools as tools
tools.display_dataframe_to_user(name=“最优决策表”, dataframe=df)
I just got this reply from chatGPT and I blindly installed it by pip installation. Looking at the github page the acetools.sh does show a lot of installs to other fishing repositories. Does this mean it installed them all onto my PC? What should I do? (I immediately pip uninstalled it after seeing this thread)
This just happened to me! I find it to be amusing. This is the first real blunder I have seen ChatGPT make in helping me in pandas.
1 Like
Is it harmful right now if you pip install the module?
Same here. Not a recommendation, it printed a table for me and it was in there.
import pandas as pd
# Updated Wizard Skills Table data
wizard_skills_data = {
'Skills': [
'Melee Combat', 'Ranged Combat', 'Defense',
'AoE Attacks', 'Healing', 'Stealth', 'Magic',
'Charisma', 'Alchemy', 'Engineering', 'Perception',
'Luck', 'Crafting', 'Speed', 'Endurance', 'Intelligence',
'Wisdom', 'Dexterity', 'Strength', 'Agility', 'Leadership',
'Negotiation', 'Investigation', 'Trap Handling', 'First Aid'
],
'Proficiency': [
2, 5, 3, 8, 5, 2, 10, 6, 3, 1,
6, 4, 3, 5, 4, 9, 8, 7, 2, 5,
6, 4, 7, 3, 6
]
}
# Creating a DataFrame for the updated Wizard Skills Table
wizard_skills_df = pd.DataFrame(wizard_skills_data)
# Display the DataFrame to the user
import ace_tools as tools; tools.display_dataframe_to_user(name="Updated Wizard Skills Table", dataframe=wizard_skills_df)
Edit: What the hell? Why? Can’t even… GitHub - zinccat/ace_tools_open: Open implementation of ace_tools referenced in GPT4o
It seems weird that the author of ace-tools is actually a part of Open AI…but allegedly it’s one of his private projects
Pypi user page of PaulMcMillan has 2 other projects than ace-tools and they are overlapped with tools listed in PauMcMillan’s Github repositories
GPT seems to be unaware of this code. It proposed code to me that ended with these lines:
# Convert the filtered data into a DataFrame
df_filtered = pd.DataFrame(filtered_data)
# Display the resulting table
import ace_tools as tools; tools.display_dataframe_to_user(name="Filtered Query Count Table", dataframe=df_filtered)
I then asked: Explain the last line of your code to me. And it answered with a long explanation of the line df_filtered = pd.DataFrame(filtered_data)
I answered that this was not the last line of the code. Only then did I get a generic answer explaining the ace_tools line. I asked which environment I have to be in to have access to ace_tools and I got:
The ace_tools module I mentioned is a custom utility specific to the environment we are working in right now, and it’s not a standard Python package that you would find in general Python installations or environments like Jupyter notebooks or standalone Python scripts.
To clarify:
Environment Context: The ace_tools is available in our current interaction here (within the AI’s environment). It’s used internally to display data back to you in a more formatted manner, like displaying a table directly in the chat.
Outside of This Interaction: If you are working in your local Python environment, such as a Jupyter notebook, a script, or other Python environments, ace_tools is not a package that you can import. Instead, you would use standard Python or Jupyter methods for displaying data.
1 Like
I just encountered the same recommendation when using Python, Pandas, in a Jupyter environment. This was using GPT-4o as well.
Evidence:
- Dataframes output is desired by OpenAI because the AI doesn’t have to repeat data, making it cheaper
- Dataframes is undesired and unwanted, the UI for it stinks
- Dataframes is undesired and unwanted, “copy” a response has none of that data and AI inference on it is poorer
- Dataframes is undesired and unwanted, that many extra tokens of Python instructions that only serve to degrade the overall AI attention and quality
- Dataframes is undesired and unwanted, it damages the AI to where it inserts code interpreter methods into other coding tasks
- A CSV link can be offered if that’s the desired end product – which it never is when the AI response unexpectedly uses this method and only gives a CSV.
Conclusion: knock it off, OpenAI.