Help LLM Training

[Oceaneyes123]

Hello po, paano po mag train ng LLM sa sariling codebase? (parang github copilot kumbaga)

 

[jccarrera]

ff

 

[alist1986]

Kailangan mo talaga ng guide dahil mahirap ipaliwanag yan sa isang upuan he he, Kaunti lang ang alam ko kundi basics. Ang importante, alamin mo muna yung process, terms,methods, conditions, saan ka kukuha ng datasets, anong hardware kailangan, anong tools yun gagamitin (online/offline), atbp. para mo ma-train yung LLM like pretraining, finetuning, reinforcement learning from human feedback (RLHF), and adapters. Hindi yan straightforward pero at least one will do for a start.
This might help as starter: You do not have permission to view the full content of this post. Log in or register now., You do not have permission to view the full content of this post. Log in or register now....
Gamit ka ng iba't ibang search engines and look for guides that you can follow using these queries: "AWESOME LLM TRAINING, GITHUB", "how to train llm using custom datasets", "GUI to train LLM, github".....Sa github/Hugginface ay maraming info, pero pagtiyagaan mong basahin lahat para maliwanag sa'yo.
Ang basics ng training ay parang magulang o teacher in real life. You talk, converse and input the right information to the child, correct his/her mistakes, evaluate his learning to see how he/she improves on and on (again and again if it is needed).
PS.
Heto, hinanapan na kita ng complete LLM dev channel. Bahala ka na dyan.You do not have permission to view the full content of this post. Log in or register now.
Kumpleto yan sa tools. bahala ka na sa datasets mo at conversion kung kailangan.