Wow it's seam instantaneous ! on what machine did you run the model ? I tried to use mistral nemo on a 4090 (using python) but the message get 5 min before being return by the model.
Couldn't get completion to work. Out of the box it uses markdown with a bunch of explanatory text. I changed the prompt, turned off multiline completion in config. Still markup and unusable. Changed to stable-code and completion stopped working completely. Deleted and reinstalled extention. It remembered model making me think it didn't really uninstall. Would be nice if I could get it to work. Thanks for vid. shell-gpt was new to me. Cheers.
@@AICodeKing is there anybody doing a video about what's the cost comparision of these 3 options: (1) use smt like copilot and pay 20-29$ per person ; (2) use some API from mistral & nvdia like you said ; (3) invest on hardware to run locally ? I think for option 3 to be viable for companies, we will have to refine the hardware spec very carefully to get the most value out of it.
I found myself watching all of your videos lol but one thing you never explain is how many resources we need to run the local llm
Are you creating a video for aider using this?
Ouch! You have done it! Great! Amazing dude! THANKS!
Finally it's here
Dancing to the intro
Wow it's seam instantaneous ! on what machine did you run the model ? I tried to use mistral nemo on a 4090 (using python) but the message get 5 min before being return by the model.
Great instruction! Btw, can you make a video about fine tune?
This plus aider would be a huge advancement
Love it thanks @AICodeKing!
This is the first-time I've made a Super-Thanks contribution to any channel. Thank you kindly!
Intro is the best 😁 welcome to another happy day with AI
How are you doing this? Its literally more than 1 high quality videos per day.
what about deepseek coder v2 ? isn't it better for coding than nemo? btw doesn't ds coder v2 get some update recently?
This channel already has review on new deepseek-coder-v2 recently. Just check it out.
Couldn't get completion to work. Out of the box it uses markdown with a bunch of explanatory text. I changed the prompt, turned off multiline completion in config. Still markup and unusable. Changed to stable-code and completion stopped working completely. Deleted and reinstalled extention. It remembered model making me think it didn't really uninstall. Would be nice if I could get it to work. Thanks for vid. shell-gpt was new to me. Cheers.
Still waiting for aider deepseek coder for editing local files
It already works , aider and deepseeker , you need add the local file with command
Wow new intro 🎉🎉
VScode refuses to actually uninstall it, keeps coming back with edited prompt. Now I have to find out how to manually kill the thing.
Have you tried Agent Zero yet?
Thank you
So in the company's perspective, do I have to provide each of my employee a GPU to run this?
No, You can get an API from Mistral or Nvidia and connect that with ContinueDev.
@@AICodeKing is there anybody doing a video about what's the cost comparision of these 3 options: (1) use smt like copilot and pay 20-29$ per person ; (2) use some API from mistral & nvdia like you said ; (3) invest on hardware to run locally ?
I think for option 3 to be viable for companies, we will have to refine the hardware spec very carefully to get the most value out of it.
Thanks!
Thanks a lot ❤️
How to make it work with ollama running on docker?
You can do that by changing the base url
“Anyway”
On windows, i am using LM Studio. can any want tell me how to integrate this with vs code.?
Yes, you can do that with ContinueDev