New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Syntax Errors Of Doom
Just got finished writing my first agent for outlook to help me with emails. Tried LLama 2, but got hosed by the syntax Errors. Ended up with LLaMa 3 I believe but, the quality of responses are not as good that I'm sure I'll get with LLama 2. Anyone have any experience with the different AI models that can clean up the syntax errors?
Thx
Comments
Claude 3 Opus
I thought you had such agent for LET.
On a serious note tho, I assume if you want self-hostable model (you mentioned llama) - you can try gemma3n, released couple days ago from Google.
Phi4 and Qwen2.5-coder/Qwen3 are also quite good at coding tasks for their size.
However I feel all those local models are very much behind current LLM pioneers (chatgpt, claude, gemini) - mostly due to size limitation.
awesome thx!
lol no, I write my own responses
Exactly on the local models. There appears....... to be a solution to make a local data base to have ChatGPT check it each time but, seems like it will take a ton of coding to get it to work the way I want it to to.
This I have a lot of information on. My company tested dozens of open source models.
We learned you have to break the work down into micro tasks with AI and engineer long complex prompts for each step. DM me if you want to collaborate.
for production only use 4.1, sonnet 4, and deepseek v3.
ok thank you definitely. At point it re-wrote the same code I would say 15 times in row with syntax error after syntax error.
awesome thank you