All your videos have been spot on - you have a good fast pace, yet cover the details/steps. You take on a decent sized problem that is interesting and inspires other ideas and see it through to completion. So they are just great for educating on the topic you cover. Great work 👍👍👍
Amazing video. I have been looking for exactly this for weeks, and since I'm not a coders, I tried Flowise too and I didn't get good results. Your video is brilliant to me. Thank you for sharing this.
Basically a flow connector for AI- Research Tweets. If enough of these are made, for each of the major information sources an agent could enhance its knowledge, then vectorize its findings and then prompt itself as it completes a task with the researched data.
Thanks, the program runs fine, I took out the query in streamlit for the OPEN_AI_Key and the script around the environment variable to OPEN_AI_Key. Then it worked great. Thanks a lot
@@AIJasonZ I made several modification: llm = OpenAI(model_name="gpt-3.5-turbo", temperature=.7, openai_api_key=OpenAI_API_KEY), and OPENAI_API_KEY = os.getenv("OPENAI_API_KEY"), openai.api_key = OPENAI_API_KEY - Then it works in my enviroment. I got this error before, if i clone and start the UI "ValidationError: 1 validation error for OpenAIChat __root__ Did not find openai_api_key, please add an environment variable `OPENAI_API_KEY` which contains it, or pass `openai_api_key` as a named parameter. (type=value_error) - I asked ChatGPT how I can fix the problem "I am not an developer, only a user"
2:32 Hey Jason, do you think that some targeted 'human in the loop' moments could help at any point in this project or any of the other projects you've been working on? Seems like there's this great push to see how much the AI can do on its own, but I wonder if any problems could be solved by us humans in the loop at the right points in time.
Hey Jason, great video. Thanks for sharing! Q: would you say this is an autonomous agent or an ai-agent with web connection and prompt-chain-instructions? I feel an autonomous agent should be able to make decisions by itself and decide which tool to use and when.
Hey Joaquin, thanks a lot for the notes! You are absolutely right, I was planning to use agent, i think next version can be using agent with custom tools: 1) find content 2) write threads. I used LLMChain now instead because it is still cost effective for now for tasks with predictable steps. (I found agent can take many unnecessary steps sometimes to complete tasks, which means it is slower and more expensive to run). But maybe it can be solved with good prompt!
Yess, it makes sense. Auto-agents are still in beta mode. Anyway, it'd be super nice to see that video with the auto-agent finding the content, summarising it and writing the threads. I imagine the key element in that case is investing time on creating a very good prompt so it doesn't get lost.
Hi! , first of all, congrats for your contents, all of them are spot on. I tried to see the repo for this video but links are not working. Some idea? Many thanks
Hey Jason. This is on fire. Pls can you share the relevance template file for this autonomous researcher. I will like to study how you did it on relevance. There's not much tutorials on relevance online. Pls 😊
Hello,, this is a very interesting video, any way to get a more in-depth tutorial, explaining from the beginning all the tools you use, some kind of a step-by-step tutorial?
Thanks for the content and sharing!, based on my experience, I don't think that GPT can actually read and article to choose the best ones.. my assumption is that it is done base on URL itself...
@@AIJasonZ a lot of limitation in terms of volumes. The problem is that I can’t find any a updated tutorial nor even an example that works with the new Twitter API.
Amazing content my friend. You did some great work here. If you ever want to collab with an old school programmer and prompt engineer, hit me up. I am very easy to find without dropping any links here!
![Noob To Max With DRAGON REWORK In Blox Fruits [FULL MOVIE]](http://i.ytimg.com/vi/LnBMOinoOvA/mqdefault.jpg)
All your videos have been spot on - you have a good fast pace, yet cover the details/steps. You take on a decent sized problem that is interesting and inspires other ideas and see it through to completion. So they are just great for educating on the topic you cover. Great work 👍👍👍
Amazing video. I have been looking for exactly this for weeks, and since I'm not a coders, I tried Flowise too and I didn't get good results. Your video is brilliant to me. Thank you for sharing this.
Thanks mate! Yea flowise & langflow is still quite limited at the moment unfortunately
Everyone is a coder nowadays with Llama or chatgpt4
Basically a flow connector for AI- Research Tweets.
If enough of these are made, for each of the major information sources an agent could enhance its knowledge, then vectorize its findings and then prompt itself as it completes a task with the researched data.
Totally agree, imagine all those agents compose on top of each other, it is kinda crazy!
Great walk through, Jason. Super helpful.
Glad you enjoyed it!
Awesome Jason
Thanks for sharing. I was trying to do this a different way but your way makes much more sense
Thanks Williams! I think there might be more effective way too like utilising agents!
Let me know what other topics you want to see too!
Great video. Great Use Case. Can't wait to try it.
Glad you like it!
Your stuff is unusual and great. Would love if future stuff included local installs w/o sign-up needs.
Thanks mate! what do you mean by local installs?
Thanks, the program runs fine, I took out the query in streamlit for the OPEN_AI_Key and the script around the environment variable to OPEN_AI_Key. Then it worked great. Thanks a lot
thanks Holger! wait - so the link didn't work with the OPEN_AI_KEY?
@@AIJasonZ I made several modification: llm = OpenAI(model_name="gpt-3.5-turbo", temperature=.7, openai_api_key=OpenAI_API_KEY), and OPENAI_API_KEY = os.getenv("OPENAI_API_KEY"), openai.api_key = OPENAI_API_KEY - Then it works in my enviroment. I got this error before, if i clone and start the UI "ValidationError: 1 validation error for OpenAIChat __root__ Did not find openai_api_key, please add an environment variable `OPENAI_API_KEY` which contains it, or pass `openai_api_key` as a named parameter. (type=value_error) - I asked ChatGPT how I can fix the problem "I am not an developer, only a user"
@@holgergelhausen8616 ah got it, yea I had those issue sometimes too, quite annoying indeed!
2:32 Hey Jason, do you think that some targeted 'human in the loop' moments could help at any point in this project or any of the other projects you've been working on? Seems like there's this great push to see how much the AI can do on its own, but I wonder if any problems could be solved by us humans in the loop at the right points in time.
Thank you Jason. Appreciate your work, really helpful!
Glad it was helpful!
Thank you Jason!
Take home message: Guess it wasnt such a good idea when you share it with the world 😅
Very nice video though ♥♥
Amazing video. How much does this cost to run? Lets see I make 20 threads a day? I'm still trying to work out what is paid for and what is free?
Another great video. Great work, thank you.
Thank you mate!
end of the day It will be only be valuable product if your using for your own value
true!
@@AIJasonZtry using gpt Engineer and the function call api
great work
Hey Jason, great video. Thanks for sharing!
Q: would you say this is an autonomous agent or an ai-agent with web connection and prompt-chain-instructions?
I feel an autonomous agent should be able to make decisions by itself and decide which tool to use and when.
Hey Joaquin, thanks a lot for the notes! You are absolutely right, I was planning to use agent, i think next version can be using agent with custom tools: 1) find content 2) write threads.
I used LLMChain now instead because it is still cost effective for now for tasks with predictable steps. (I found agent can take many unnecessary steps sometimes to complete tasks, which means it is slower and more expensive to run). But maybe it can be solved with good prompt!
Yess, it makes sense. Auto-agents are still in beta mode. Anyway, it'd be super nice to see that video with the auto-agent finding the content, summarising it and writing the threads. I imagine the key element in that case is investing time on creating a very good prompt so it doesn't get lost.
Hi! , first of all, congrats for your contents, all of them are spot on. I tried to see the repo for this video but links are not working. Some idea? Many thanks
Great video! Keep it up man 👍👍
Thanks mate, you too! Awesome videos :)
Hi, great video, you should also create a video on SuperAgi
Thanks mate! I will look into it. :)
Amazing
Great video. Thanks a lot
the demo is not working anymore :(
is langchain easy to learn?
the link to the code seems to be broken...:)
I don't understand why did you remove the project from github ;(
Hey Jason. This is on fire. Pls can you share the relevance template file for this autonomous researcher. I will like to study how you did it on relevance. There's not much tutorials on relevance online.
Pls 😊
great video, are you going to upload the repository on GitHub or huggingface?
Hello,, this is a very interesting video, any way to get a more in-depth tutorial, explaining from the beginning all the tools you use, some kind of a step-by-step tutorial?
Thank you!
Would be great with subtitles and if you zoom in when showing code 🙏
Thanks for the feedback mate, i will improve those in the future videos ❤️
which huggingface model would you use instead of openai for this?
I haven’t found one is that very close to openai tbh, but falcon seem to be the best one now
great video Jason. I hope that was not your Key at the end. Just a heads up.
oh shot.... thanks so much man! i just reset the Key 🙏
@@AIJasonZ Yeah it would have happend to me too.
Thanks for the content and sharing!, based on my experience, I don't think that GPT can actually read and article to choose the best ones.. my assumption is that it is done base on URL itself...
Yea it can’t read article directly, but we add a function to do web scrape to get content as an extract step :)
Super interesting. BTW is there anyone here that knows how to use the API V2 to post on Twitter? (and even better to search) ???
I’d love to know that too, heard v2 api is quite different with some limitation!
@@AIJasonZ a lot of limitation in terms of volumes. The problem is that I can’t find any a updated tutorial nor even an example that works with the new Twitter API.
Your GH repo is showing a 404.
you must set temperature to 0 or it will generate little bit false info
Amazing content my friend. You did some great work here. If you ever want to collab with an old school programmer and prompt engineer, hit me up. I am very easy to find without dropping any links here!
ohh, thank you so much Mark! let me know if there is any other topic you are interested in but coulnd't find any good content about!
Sent you an email mate :)
太牛逼了woc，你自己读文档学的么？
Hey Jason great tutorial, can I contact you for business?
Sure thing! Email me pls! Jason.Zhou.design@gmail.com
Great video! subbed to support