rasbt / LLMs-from-scratch

Implement a ChatGPT-like LLM in PyTorch from scratch, step by step
https://www.amazon.com/Build-Large-Language-Model-Scratch/dp/1633437167
Other
32.85k stars 3.95k forks source link

ch06/03 fixes #336

Closed d-kleine closed 2 months ago

d-kleine commented 2 months ago
rasbt commented 2 months ago

Thanks for the update. Actually, it might be shorter and sufficient to use args.bert_model (updated it in the PR). But good callout!

d-kleine commented 2 months ago

@rasbt Thanks - yeah, I agree, using --model is more straightforward :)

What about the train_bert_hf_spam.py and sklearn-baseline.ipynb files, are they still needed in this folder? For me, it seems like they don't have any purpose in the code for this project anymore.

rasbt commented 2 months ago

Regarding those files, I agree, they are not used in this folder here, but I am preparing a post with some comparisons, and I needed them there, so I'd like to keep them there for reference.

d-kleine commented 2 months ago

Okay thanks