forgot to mention in the video: you will need to start the Ollama app on your PC before running the program. also made a mistake with 2 of the colorized print statements: the print statement i put on line 113 in the video, should be moved to line 102. then on line 113 print: “DATA SCRAPED FROM: {page_link}”
I followed along with you step-by-step and my line 113 looks like this: "context = page_text" as part of the ai_search while loop. Line 102 currently reads as "best_result = best_search_result(s_results=search_results, query=search_query) in the same while loop. Things shifted around during the process of drafting and so while in the video it might have been on line 113 and 102 at the time, the finished product by the end of the video doesn't align with those reference points. It could be, however, that I have no idea what I'm talking about. It just looks to me like "DATA SCRAPED FROM: {page_link}" doesn't fit on 113.
For some reason, when I run this code and talk to the model, it tells me that it has access to the internet, but I ask for the next ufc event date and it tells me 2023. So its gaslighting me about being able to see search results lol how do I fix that?
figure out if it's searching or not. if searching when it should, figure out what link it's selecting. modify system messages and or try a different language models until you get it to perform correctly. these smaller language models are going to hallucinate commonly. If you really want to maximize their performance for each of the specific tasks in the program, i would fine-tune a model for each task. basically, if you wanna improve upon these python programs, i would think through the code top to bottom of where my issue may lie in the program
I would also play around with the model temperature as well. I get a large variety of different results from the same models, simply by changing the temp 1/10th of a point.
hello bro please next video soon and also make one for gemini latest update and please include it's all features and also add it's function calling pls
i don't give microsoft my code anymore. if you guys need it, watch my video long enough to know how to access my source code. because i don't make videos for people to click off and yoink my code without supporting me continuing to make these programs you guys seem to want😉
@@Ai_Austin man, I'm just here to yoink some code :P Gotta say... your method of making me watch the video and type it all out myself is actually a lot more beneficial in the long-run because you're actually walking me through the coding process. As a complete novice, I really appreciate it.
there is nothing wrong with for profit. it's the deception that i am against. it's the calling yourself "openai" and being the most closed and biggest company in ai. it's the selling your product as greater than it really is that i am against. and it's the control of the most powerful tool humans will create i am against. i know my morals pal, but if you think you can do what i do while making $0 to feed yourself for the thousands of hours your put into it, please help out in getting more humans to know these skills. i can't do that but would totally respect if you can invest thousands of hours into something (without being a billionaire) and make $0 for your time, while sustaining your ability to keep putting time into helping humans. thanks for your opinions bud
Thank you for sharing this. In my case I am receiving a captcha page as a response from duckduckgo, asking to prove of If am human. Can you please suggest a workaround to resolve this?
forgot to mention in the video:
you will need to start the Ollama app on your PC before running the program.
also made a mistake with 2 of the colorized print statements:
the print statement i put on line 113 in the video, should be moved to line 102.
then on line 113 print:
“DATA SCRAPED FROM: {page_link}”
I followed along with you step-by-step and my line 113 looks like this:
"context = page_text" as part of the ai_search while loop.
Line 102 currently reads as "best_result = best_search_result(s_results=search_results, query=search_query) in the same while loop.
Things shifted around during the process of drafting and so while in the video it might have been on line 113 and 102 at the time, the finished product by the end of the video doesn't align with those reference points.
It could be, however, that I have no idea what I'm talking about. It just looks to me like "DATA SCRAPED FROM: {page_link}" doesn't fit on 113.
The way you explained each step and reason for each piece of code so fluently is exactly the kind of guidance n00bs like me need.
Just another incredible video. Y’re the best !!
I support you! Thank you for this video.
:)
Does yours already come with a system to identify whether it's for searching or just a "normal" prompt? If not, can you teach how to add it?
i bet if you watch you'll find out
@@Ai_Austin How do I get him to cite the sources used, like just the link, without an icon, etc.?
i would just print the link selected by the best_search_result function
Great video with explanation. Really can't learn without this guidance!.
fight against the for profit AI overlords with open source, locally hosted and privacy as a right by default. i dont trust sam or elon or zack!
great i need more learn about make local ai assistant can do
Can u show how to train our data from api with ollama
Thanks man for the great tutorial. i'm just having 1 issue. whenever ollama responds, it adds 2 empty lines after each word. the print('
') makes it do that i believe. btw, it does respond and all... but as i said, 2 blank lines after each word from the response. is it just me ?
For some reason, when I run this code and talk to the model, it tells me that it has access to the internet, but I ask for the next ufc event date and it tells me 2023. So its gaslighting me about being able to see search results lol how do I fix that?
figure out if it's searching or not.
if searching when it should, figure out what link it's selecting.
modify system messages and or try a different language models until you get it to perform correctly.
these smaller language models are going to hallucinate commonly. If you really want to maximize their performance for each of the specific tasks in the program, i would fine-tune a model for each task.
basically, if you wanna improve upon these python programs, i would think through the code top to bottom of where my issue may lie in the program
I would also play around with the model temperature as well. I get a large variety of different results from the same models, simply by changing the temp 1/10th of a point.
hello bro please next video soon and also make one for gemini latest update and please include it's all features and also add it's function calling pls
wheres the github link for this???????????????
whats a github link?
@@Ai_Austingithub repository I suppouse he wanted to say.
i don't give microsoft my code anymore. if you guys need it, watch my video long enough to know how to access my source code. because i don't make videos for people to click off and yoink my code without supporting me continuing to make these programs you guys seem to want😉
@@Ai_Austin man, I'm just here to yoink some code :P Gotta say... your method of making me watch the video and type it all out myself is actually a lot more beneficial in the long-run because you're actually walking me through the coding process. As a complete novice, I really appreciate it.
lol who is this sam, anyways nice video ai austin（￣︶￣）
21:12 this is very frustrating!
Woy tin ini kah TH-cam nya
"Stop paying these for-profit providers. Join my patreon today". Too bad youngsters hate themselves for profiting.
there is nothing wrong with for profit. it's the deception that i am against. it's the calling yourself "openai" and being the most closed and biggest company in ai. it's the selling your product as greater than it really is that i am against. and it's the control of the most powerful tool humans will create i am against. i know my morals pal, but if you think you can do what i do while making $0 to feed yourself for the thousands of hours your put into it, please help out in getting more humans to know these skills. i can't do that but would totally respect if you can invest thousands of hours into something (without being a billionaire) and make $0 for your time, while sustaining your ability to keep putting time into helping humans.
thanks for your opinions bud
I f love u
I'm not sure how Sam Altman exactly is stopping you from doing better. Stop moaning and Just Do It. If you think you've got the stuff.
i don't think anyones moaning here but you. we're just some chill guys building stuff. keep your negativity in your head please
Thank you for sharing this. In my case I am receiving a captcha page as a response from duckduckgo, asking to prove of If am human. Can you please suggest a workaround to resolve this?