Gorgeous! Exactly what I was in need of! Thank you so much, Mervin! I wondered, what is the way to ask Llama 3, and your video appeared. 😎 - And yes, AMAZING! I’ll use it with LM Studio instead of groq.
hey Mervin, just wanna say, huge thanks. its amazing. In future, if you manage to get some time, please create/share a video similar to this but change last part of webui to whatsapp chat bot....(with intent of support agent) i have so much more long list, but ideally if it can change multiple lanaguges that would be super awesome.......
Great video! Just wondering if similar function calling is possible through ollama web ui? I'm currently using llama 3 with with web ui and serving it through ngrok to use on phone and already loving it for most of my use cases but really want to implement function calling for particular use cases could you please let me know if something like that would be possible or to enable internet access to all the queries like if I ask it for weather updates or todays news highlights then it can do so?
If I give it a task that requires it to run multiple functions. What do you think is the most reliable way to achieve this? My current implementation is a loop. It calls 1 function each iteration and loops. Each time aware of which function it has requested to be called in previous iterations. When it is satisfied that no more functions need to be called, it calls an end function. So if I were to do that with this approach I guess I would add the end function in the "tools" with the description run this when no more functions need to be run. The user prompt would need to provide the functions called on previous iterations. And the system prompt should instruct to only call 1 function at a time? Please let me know how you would achieve this, thanks.
now i also have more questions....after finished watching... Q. can this be done with local llama8b installed via ollama.??? if so, what steps do i need to change.??
Hi there, thanks for the video! ! I'm currently grappling with an issue in the llama3 model and could really use your expertise. We've set up a weather information function, but it's hit-or-miss when it comes to recognizing and activating in response to user prompts. While this function operates seamlessly with the gpt-4-turbo model, llama3 struggles, often providing inaccurate data or none at all. This is despite the function being clearly defined with all necessary parameters, suggesting that the model's trigger mechanism might be at fault. Could you shed some light on how to enhance llama3's consistency with function activation? Thanks!
Hi Mervin, the groq rate limit is a big issue for my use case, can i use this same method to use another similar hosted llama 3 70b with crewai like openrouter api or can any api be used instead of groq with your method using web scraping?
This is cool. But (forgive the n00b questions) I’m unfamiliar with ‘groq’, I thought that was Elon Musk’s LLM? presumably it’s a library for using ‘tools’. And this is an illustrative example, the ‘tool’ could be anything, an executable, for example. Otherwise why not just use a RAG for this particular example, given that LLMs are unreliable/unpredictable?
Groq is different from Grok Grok is Elon Musk’s LLM. Here we are discussing about Groq which provides faster response when we ask a question. RAG generally used for unstructured text data. In this video tutorial the data is structured and hence we are using tools instead of RAG to be more efficient.
@@MervinPraison Thanks Mervin. BTW love your concise code snippets, they are *very* helpful. Ollama is gaining traction, fast, but the documentation and other content providers were still focused on Open AI, Langchain, and associated integrations until only a few weeks ago, so your coverage has been helpful.
Any specific area of the video you wish to zoom out? Because I used various zoom level for each part of the video. Just so that next time it is easy for you to view.
Thank you for sharing this!
Thank you :)
@@MervinPraison , I wish I could do more! ❤️
with groq's speed this could be extremely powerful if implemented properly, thanks for the video
Thank you
Very clear, thanks!
Very cool! Need to see how to implement document embedding with this setup you shared.
Absolutely brilliant! THank you
Thank you
Gorgeous! Exactly what I was in need of! Thank you so much, Mervin! I wondered, what is the way to ask Llama 3, and your video appeared. 😎 - And yes, AMAZING! I’ll use it with LM Studio instead of groq.
NIce :) Glad it was helpful.
Wow. This is what I was looking for. Thank you so much
Short and concise. Thank you
hey Mervin, just wanna say, huge thanks. its amazing.
In future, if you manage to get some time, please create/share a video similar to this but change last part of webui to whatsapp chat bot....(with intent of support agent)
i have so much more long list, but ideally if it can change multiple lanaguges that would be super awesome.......
Hi, Mervin. Thank you for your excellent presentation and tutorial. Could you please perform the procedures in Docker Compose?
Great video! Just wondering if similar function calling is possible through ollama web ui? I'm currently using llama 3 with with web ui and serving it through ngrok to use on phone and already loving it for most of my use cases but really want to implement function calling for particular use cases could you please let me know if something like that would be possible or to enable internet access to all the queries like if I ask it for weather updates or todays news highlights then it can do so?
Very nice tutorial. Well done and thank you
Thanks for sharing Mervin! I need your help to implement such a tool in WordPress pages.
If I give it a task that requires it to run multiple functions. What do you think is the most reliable way to achieve this? My current implementation is a loop. It calls 1 function each iteration and loops. Each time aware of which function it has requested to be called in previous iterations. When it is satisfied that no more functions need to be called, it calls an end function.
So if I were to do that with this approach I guess I would add the end function in the "tools" with the description run this when no more functions need to be run. The user prompt would need to provide the functions called on previous iterations. And the system prompt should instruct to only call 1 function at a time?
Please let me know how you would achieve this, thanks.
now i also have more questions....after finished watching...
Q. can this be done with local llama8b installed via ollama.??? if so, what steps do i need to change.??
Fantastic stuff. Going to give this a go myself!
Thank you
"this is amazing"🎉
Thank you
Hi there, thanks for the video! ! I'm currently grappling with an issue in the llama3 model and could really use your expertise. We've set up a weather information function, but it's hit-or-miss when it comes to recognizing and activating in response to user prompts. While this function operates seamlessly with the gpt-4-turbo model, llama3 struggles, often providing inaccurate data or none at all. This is despite the function being clearly defined with all necessary parameters, suggesting that the model's trigger mechanism might be at fault. Could you shed some light on how to enhance llama3's consistency with function activation? Thanks!
Great short concise video. Thanks for making it. Looking at the source code link, I noticed that ui.py is not listed. Could you update. Thanks
Added now :) Thanks for the reminder
What if not all the query needs to call the tool?
This is not only on stuff from the top of the AI wave, but the crème de la crème
Mervin
Thank you 🙏
Hi Mervin, the groq rate limit is a big issue for my use case, can i use this same method to use another similar hosted llama 3 70b with crewai like openrouter api or can any api be used instead of groq with your method using web scraping?
What if I want to use llama locally or on my own host?
how to get migrste from Groq API to my own?
Awesome stuff. Really useful.
Thank you
please let me know if i finetune llama-3-8b-instruct model in my local mechine what are the system specifications?
My RTX 3050 is fine
Any reason you don't put these on github so people can run them and play with them?
It’s in the description
@@MervinPraison i see a link to your blog but no link to github even when i search this whole page?
This is cool. But (forgive the n00b questions) I’m unfamiliar with ‘groq’, I thought that was Elon Musk’s LLM? presumably it’s a library for using ‘tools’. And this is an illustrative example, the ‘tool’ could be anything, an executable, for example. Otherwise why not just use a RAG for this particular example, given that LLMs are unreliable/unpredictable?
Groq is different from Grok
Grok is Elon Musk’s LLM.
Here we are discussing about Groq which provides faster response when we ask a question.
RAG generally used for unstructured text data. In this video tutorial the data is structured and hence we are using tools instead of RAG to be more efficient.
@@MervinPraison Thanks Mervin. BTW love your concise code snippets, they are *very* helpful. Ollama is gaining traction, fast, but the documentation and other content providers were still focused on Open AI, Langchain, and associated integrations until only a few weeks ago, so your coverage has been helpful.
Can you make a video usin Crew Ai?
cant we use gro wtihotu API_key?
plese guide anyone
i mean grog. client = Grog(api_key.....)
I tried the same code with locally running llama3 but it didn't trigger the function calling any idea why?
Llama3 8B is not as effective as 70b in regards to function calling
could you zoom out a little.
Any specific area of the video you wish to zoom out?
Because I used various zoom level for each part of the video. Just so that next time it is easy for you to view.
Groq and Grok very confusing .. latter is Elons.