[HN Gopher] How to think about creating a dataset for LLM fine-t...
___________________________________________________________________
How to think about creating a dataset for LLM fine-tuning
evaluation
Author : sebg
Score : 109 points
Date : 2024-06-27 10:29 UTC (12 hours ago)
(HTM) web link (mlops.systems)
(TXT) w3m dump (mlops.systems)
| msp26 wrote:
| For tasks like data extraction, are people doing full finetunes
| or training a LoRA? Is it any different for classification?
| clarionbell wrote:
| Unless you have GPUs available, LoRa is the only accessible
| option. That being said, if your task is simple enough, you can
| skip the problem entirely and just pick a small base model.
| swalsh wrote:
| LoRA is perfect for that. Where it fails is when you want to
| teach a new domain.
| avgbusinessuser wrote:
| great series of posts, i went down a similar path recently for a
| slightly different use case - i did not use axolotl though, i was
| worried about missing out on understanding some details due to
| potential abstractions. it's great to see documentation on how
| others tackle similar problems, i documented the process i went
| through here - https://atredis.com/blog/2024/6/3/how-to-train-
| your-large-la...
| strickvl wrote:
| There's a ton of abstraction in axolotl, for sure, but so far I
| haven't found that it gets in the way. The main competitor in
| that space seems to be Unsloth, but that only works with a
| single GPU machine, so didn't fit my purposes. I'll dive into
| your blogpost. Thanks for posting!
| avgbusinessuser wrote:
| I used unsloth, I was only using a single GPU for testing -
| looking forward to follow up posts.
| hinkley wrote:
| When you get good enough at filtering the dataset for training,
| do you still need an AI, or do you understand the problem domain
| and can use a deterministic system?
___________________________________________________________________
(page generated 2024-06-27 23:01 UTC)