Great video. The mapping is particularly impressive. I have seen a few Code Interpreter videos with mapping, but I haven't been able to do it myself. If I ask it to draw a basic map of the USA, it says that it can't but that it could guide me through the process of generating it using Python libraries like Matplotlib and Basemap. When I ask what about using Folium, it says sure it could use the Folium library, but that it can't run this code in the current environment as it does not support interactive visualisation. Just wondering if you had to do anything else to get it working? or perhaps OpenAI has ChatGPT environments at different stages in different countries? (I am in Australia).
I've noticed it does this a lot for me(in America). I think it's just a code interpreter thing. It seems like they directly used GPT-4 as the underlying model for code interpreter as opposed to training or fine-tuning a new model. It seems to 'believe' that it has the same limitations as plain GPT-4
Agree with @jacoballessio5706's point that Code Interpreter believes it has the same limitations as GPT4. I didn't have to tweak anything regarding the example in the video; but most likely then because the wind turbine CSV file had geolocation data, which CI probably picked up and then initiated the folium.
Indeed, we are quite excitable. Especially compared to Europeans. These are of course generalizations, but having lived outside of the US for the past 15 years, I have myself noticed and heard from many others that Americans (and Canadians) tend to exaggerate, easily get excited, and express enthusiasm in ways that may seem over the top for people of other nationalities.
Great job. Love Code Interpreters potential and your energy!!
Great video. The mapping is particularly impressive. I have seen a few Code Interpreter videos with mapping, but I haven't been able to do it myself. If I ask it to draw a basic map of the USA, it says that it can't but that it could guide me through the process of generating it using Python libraries like Matplotlib and Basemap. When I ask what about using Folium, it says sure it could use the Folium library, but that it can't run this code in the current environment as it does not support interactive visualisation. Just wondering if you had to do anything else to get it working? or perhaps OpenAI has ChatGPT environments at different stages in different countries? (I am in Australia).
I've noticed it does this a lot for me(in America). I think it's just a code interpreter thing. It seems like they directly used GPT-4 as the underlying model for code interpreter as opposed to training or fine-tuning a new model. It seems to 'believe' that it has the same limitations as plain GPT-4
Agree with @jacoballessio5706's point that Code Interpreter believes it has the same limitations as GPT4. I didn't have to tweak anything regarding the example in the video; but most likely then because the wind turbine CSV file had geolocation data, which CI probably picked up and then initiated the folium.
I don't really understand the "North America" comments or jokes. Are North Americams supposed to be very excitable?
Indeed, we are quite excitable. Especially compared to Europeans. These are of course generalizations, but having lived outside of the US for the past 15 years, I have myself noticed and heard from many others that Americans (and Canadians) tend to exaggerate, easily get excited, and express enthusiasm in ways that may seem over the top for people of other nationalities.