Hello eveyone , cole hi
So here i am again .
After the release of deep seek v3 , i was thrilled to know it has the same coeing power of claude and o1
So went to yestbit and the same problem ( with gpt versions ) was there , which is the LLMs being 2023 updated , and not 2024
So errors happened and i kind of believe the model wasnt trained the best with mql5 language as it makes unusual errors
So the question here is , can i make a chat ui interface ( not only UI ) but also a place on my pc to link deeps seek 3 into it
This place features the fact that it is dedicated for mql5 language ( i dont know how but maybe the keyword to it is fine tuning )
Certainly i dont mean to have deep seek 3 locally , what i mean is that can i create some kind of a fork (like bolt.diy) then link it to an api of deep seek 3 ( closest to best in cosing solving power )
Then the fact that this local or fork on my pc can make the LLM write a neat coee without errors and making sure ro delieve the right wanted logic for the code
It is a wild idea for me and it might be even wrong , i dont really know
If that is possible in anyway , can you please tell me how . Maybe in small words or extensive steps towards achieving it
Now you might propose finetuning a local LLM for mql5 , i know about this approach but here i am trying to use deep seek 3 with an api and maybe ,aybe and propably , there is a way to make it correct code ( it really messes the syntax and functions and variables badly )
But as i saw it has very good reasoning and it gets the idea ( the ultimate full idea ) wuite easily
Thanks everybody