Continue
Dev tools
ai coding assistant
SHOW HN2023
How HN took it
3 positive, 13 neutral or asking, 0 mixed, 4 critical, out of 20 comments about the project. Thread score -0.05.
Performance: 1 for, 1 againstExecution: 0 for, 2 againstIdea: 1 for, 0 againstPricing: 0 for, 1 againstDocs: 1 for, 0 against
The thread, comment by comment
Each row is a top-level comment and its replies, in HN's order. Select a cell to open that comment on HN.
positive neutral or mixed critical about another commenter or something else the maker
What HN said
What people liked
Local Code Llama via Ollama is capable and cheap to run
From 2 comments
not true anymore. give codellama-7b-instruct a try. Just install Ollama. Pretty mind-blowing performance for the ram it uses and how fast it is. It's in the ballpark of…
jasonjmcghee on HNUniteAI (https://github.com/freckletonj/uniteai) I think fits the bill for you. This is my project, where the goal is to Unite your AI-stack inside your editor (so,…
freckletonj on HN
Excitement about open code models
From 1 comment
I just love this so much. It took what, maybe 6 hours for these model binaries to be publicly available? Never seen so much excitement in tech.
ramesh31 on HN
Objections
Running a capable model locally needs expensive hardware
From 2 comments
This would be great if I had a beefy machine with 32GB ram and 16 cores. The alternatives presented are to use either Together or Standard api's to run llama-code. Both have a…
say_it_as_it_is on HNbut for code llm to be useful , the local machine need to have some very powerful GPU
yangcheng on HN
Local models may lag GPT-4 in quality
From 2 comments
Why is this post flagged? I've yet to try code stuff with AI (even Copilot). How well do local models like Code Llama work compared to GPT? When I use ChatGPT, GPT 3.5 feels a…
gettodachoppa on HNYou can't run anything even remotely close to GPT-4 locally. Large enterprises can run a private instance of it, but OpenAI is paid for that (indirectly via Microsoft).
int_19h on HN
Not yet available on Windows or Linux
From 2 comments
I don't have a Mac. So it's either have a Mac or don't run locally if you want to use this service?
Xiol32 on HNSorry about that. We are working on building the Windows and Linux versions. May I ask which specific OS you are on?
mchiang on HN
Requests
JetBrains plugin
From 1 comment
I wish someone would make an IntelliJ/Android Studio plugin for Code Llama (or another local llm). I know both platforms have their own AI features but that involves sending my…
smith7018 on HN
Linux and Windows support
From 1 comment
Not OP, but would be great to have it working on Linux. How can one help?
hawk_ on HN
Questions
How well does Code Llama compare to GPT-4?
From 1 comment
Why is this post flagged? I've yet to try code stuff with AI (even Copilot). How well do local models like Code Llama work compared to GPT? When I use ChatGPT, GPT 3.5 feels a…
gettodachoppa on HN
Why are two external services needed if running locally?
From 1 comment
Why is it necessary to sign up for two external services if this is running the LLM locally?
Xiol32 on HN
Alternatives named
UniteAI
From 1 comment
UniteAI (https://github.com/freckletonj/uniteai) I think fits the bill for you. This is my project, where the goal is to Unite your AI-stack inside your editor (so,…
freckletonj on HN
Refact
From 1 comment
at Refact we have Jetbrains plugin that works with a bunch of local code models https://github.com/smallcloudai/refact/
kateklink on HN
Cursor
From 1 comment
I don't see how things like Cursor[0] (a VS Code fork with a nicer Codex UI) have a moat or chance of profitability in the face of things like this. If you can run models locally…
dcre on HN
Sourcegraph Cody
From 1 comment
I guess I've been summoned. We're primarily focusing on VSCode, IntelliJ and Neovim for Cody. Of course I'll be working on an Emacs version, but that's kinda best-effort for now.…
_steve_yegge_ on HN
GitHub Copilot
From 1 comment
If you run Ollama locally, the cost is zero. With Replicate, you pay about $0.005 per second, so with an average of say 15 seconds per request to generate the full response, and…
sestinj on HN