31–40 of 117 posts
I wonder what's the reason.
I'm new to this whole area and feeling a bit lost. How are people setting up these small LLMs like Yi-Coder locally for tab completion? Does it work natively on VSCode? Also for the cloud models apart from GitHub Copilot, what tools or steps are you all using to get them working on your projects? Any tips or resources would be super helpful!
This is easy to get "working" but difficult to configure for specific tasks due to docs being lacking or contradictory.
The difference between (A) software engineers reacting to AI models and systems for programming and (B) artists (whether it's painters, musicians or otherwise) reacting to AI models for generating images, music, etc. is very interesting. I wonder what's the reason.
You also need a knowledge of code to instruct an LLM to generate decent code, and even then it's not always perfect.
Meanwhile plenty of people are using free/cheap image generation and going "good enough". Now they don't need to pay a graphic artist or a stock photo licence
Any layperson can describe what they want a picture to look like so the barrier to entry and successful exit is a lot lower for LLM image generation than for LLM code generation.
It would be good if LLMs were somehow packaged in an easy way/format for us "novice" (ok I mean lazy) users to try them out. I'm not interested so much with the response time (anyone has a couple of spare A100s?), but it would be good to be able to try out different LLMs locally.
It would be good if LLMs were somehow packaged in an easy way/format for us "novice" (ok I mean lazy) users to try them out. I'm not interested so much with the response time (anyone has a couple of spare A100s?), but it would be good to be able to try out different LLMs locally.
It would be good if LLMs were somehow packaged in an easy way/format for us "novice" (ok I mean lazy) users to try them out. I'm not interested so much with the response time (anyone has a couple of spare A100s?), but it would be good to be able to try out different LLMs locally.
You can also use several GPU options, but they are not as easy to get working.
It would be good if LLMs were somehow packaged in an easy way/format for us "novice" (ok I mean lazy) users to try them out. I'm not interested so much with the response time (anyone has a couple of spare A100s?), but it would be good to be able to try out different LLMs locally.
The difference between (A) software engineers reacting to AI models and systems for programming and (B) artists (whether it's painters, musicians or otherwise) reacting to AI models for generating images, music, etc. is very interesting. I wonder what's the reason.
If they can take a Jira ticket, debug the code, create a patch for a large codebase and understand and respect all the workarounds in a legacy codebase, I would have a problem with it.
It would be good if LLMs were somehow packaged in an easy way/format for us "novice" (ok I mean lazy) users to try them out. I'm not interested so much with the response time (anyone has a couple of spare A100s?), but it would be good to be able to try out different LLMs locally.
Claude 3.5 Sonnet still holds the LLM crown for code which I'll use when wanting to check the output of the best LLM, however my Continue Dev, Aider and Claude Dev plugins are currently configured to use DeepSeek Coder V2 236B (and local ollama DeepSeek Coder V2 for tab completions) as it offers the best value at $0.14M/$0.28M which sits just below Claude 3.5 Sonnet on Aider's leaderboard [1] whilst being 43x cheaper…
43x cheaper is good, but my time is also worth money, and it unfortunately doesn't bode well for me that it's stumped by the first problem I throw at it.