Chapters and links and key takeaways in blog form: 00:00 - Intro: Times of change, will we stop coding? 00:56 - Presentation I made for colleagues. Chronicles of making ChatGPT write code 01:30 - 2022 I don't want to read code that does not work 2:36 - January 2023: LLMs are a bad tool for the job 3:46 - I decided to give GPT3.5 a TDD task 4:57 - Comparing GPT3.5 to human coder, is that fair? 6:15 - February-March: GPT4, Plugins, Code Interpreter, Internet access 7:42 - GPT4 with Code Interpreter finds data entry error that Data Journalism students don't 9:18 - Testing GPT4 on TDD task, I was very surprised! Result react project: codesandbox.io/p/devbox/ch2rx2 12:36 - May: LLMs making and testing tools for themselves to solve problems Voyager paper: arxiv.org/abs/2305.16291 14:48 - Experiment with ChatGPT Browser JavaScript interpreter www.wonderwhy-er.com/ChatGPTJ... / supercharge-your-chatgpt-conversations-wit... 18:56 - Changing views. What can't it do if it can write code and execute code? 19:48 - July: ChatGPT Code Interpreter, incredible but limited / asking-openai-advanced-data-analysis-to-an... 24:59 - Gorrila and TollLLM, LLMs that know when and how to use 16 thousands of APIs arxiv.org/abs/2305.15334 arxiv.org/abs/2307.16789 26:44 - August Experiment: ChatGPT Desktop Commander github.com/wonderwhy-er/ChatG... 29:30 - September, Recap of 2023 so far, wild ride 31:32 - What's next? Exploring more and building ChatGPT Server Commander in public github.com/wonderwhy-er/ChatG... 33:26 - What did I learn on limits of LLMs since September. Small recap 33:53 - Other plans for next videos, subscribe if interested! Key takeaways in blog form: wonderwhy-er.medium.com/chronicles-of-making-chatgpt-write-code-4f83b3d09d33 www.linkedin.com/pulse/chronicles-making-chatgpt-write-code-eduard-ruzga-gnh5f/?trackingId=7OqJoHqITt6evLZWR9HUig%3D%3D
Chapters and links and key takeaways in blog form:
00:00 - Intro: Times of change, will we stop coding?
00:56 - Presentation I made for colleagues. Chronicles of making ChatGPT write code
01:30 - 2022 I don't want to read code that does not work
2:36 - January 2023: LLMs are a bad tool for the job
3:46 - I decided to give GPT3.5 a TDD task
4:57 - Comparing GPT3.5 to human coder, is that fair?
6:15 - February-March: GPT4, Plugins, Code Interpreter, Internet access
7:42 - GPT4 with Code Interpreter finds data entry error that Data Journalism students don't
9:18 - Testing GPT4 on TDD task, I was very surprised!
Result react project: codesandbox.io/p/devbox/ch2rx2
12:36 - May: LLMs making and testing tools for themselves to solve problems
Voyager paper: arxiv.org/abs/2305.16291
14:48 - Experiment with ChatGPT Browser JavaScript interpreter
www.wonderwhy-er.com/ChatGPTJ...
/ supercharge-your-chatgpt-conversations-wit...
18:56 - Changing views. What can't it do if it can write code and execute code?
19:48 - July: ChatGPT Code Interpreter, incredible but limited
/ asking-openai-advanced-data-analysis-to-an...
24:59 - Gorrila and TollLLM, LLMs that know when and how to use 16 thousands of APIs
arxiv.org/abs/2305.15334
arxiv.org/abs/2307.16789
26:44 - August Experiment: ChatGPT Desktop Commander
github.com/wonderwhy-er/ChatG...
29:30 - September, Recap of 2023 so far, wild ride
31:32 - What's next? Exploring more and building ChatGPT Server Commander in public
github.com/wonderwhy-er/ChatG...
33:26 - What did I learn on limits of LLMs since September. Small recap
33:53 - Other plans for next videos, subscribe if interested!
Key takeaways in blog form:
wonderwhy-er.medium.com/chronicles-of-making-chatgpt-write-code-4f83b3d09d33
www.linkedin.com/pulse/chronicles-making-chatgpt-write-code-eduard-ruzga-gnh5f/?trackingId=7OqJoHqITt6evLZWR9HUig%3D%3D
Thank you for the information, Eduards! 👍
Informative review, everything is to the point. Thanks!
Love it!