@rasbt

–

01 August 2023 1h27

it's of course all about LLMs 😛

1 August 2023

Suggestions for More Interesting Model Training Progression Sequences

By

–

31 July 2023 22h08

Ok fair, it would have been of course more interesting if they did SFT v1 -> SFT v2 -> SFT v3 -> … SFT v1 -> SFT v2 -> RLHF v1 -> … RLHF v1 -> RLHF v2 -> RLHF v2 -> …

DPO vs RLHF: Genuine Competition or Surface-Level Improvement?

By

–

31 July 2023 16h22

Yeah exactly. I wonder if DPO genuinely competes with RLHF or if the models only looks good on the surface but are worse under closer inspection (like with imitation models)

Human Evaluations Included in Upper Right Instead of GPT-4

By

–

31 July 2023 16h21

I included human evaluations in the upper right as opposed to GPT-4 evaluations

Llama 2 Paper Analysis: Supervised Finetuning vs RLHF Evaluation

By

–

31 July 2023 14h51

Finally got a chance to spend some time with the 77-page Llama 2 paper (
https://
arxiv.org/abs/2307.09288). Here are some takeaways at a glance. Appreciate that someone finally did a comprehensive supervised finetuning vs RLHF evaluation! (Lower right)

InstructGPT and ChatGPT represent significant advancement beyond previous models

By

–

31 July 2023 12h59

I’d say that InstructGPT, ChatGPt, and everything after (which were all released after this paper) were quite a leap forward

Fix for torchtext dataset class labels in machine learning book

By

–

30 July 2023 16h54

Thanks! I think the main issue was that newer versions of torchtext apparently changed the class labels in that dataset. It should also be fixed in the original repo here now: https://
github.com/rasbt/machine-
learning-book/blob/main/ch15/ch15_part2.ipynb
… Sorry for the hassle!

30 July 2023

Training efficiency boost through compressed data caching optimization

By

–

30 July 2023 14h46

I haven't! Thank for sharing! But a note on the efficiency: caching the compressed train lengths makes absolute sense, the speed up is approx 3.8 sec/iter -> 3.1 sec/iter

30 July 2023

Neural Networks and Gzip versus Large Language Models Deep Dive

By

–

30 July 2023 14h21

Following up on some discussions of the "NN + Gzip versus LLMs" methods a few weeks back here's my little deep dive with a reimplementation and additional experiments: https://
magazine.sebastianraschka.com/p/large-langua
ge-models-and-nearest/
… This includes fixing the tie-breaking and more!

30 July 2023

Using PyTorch Vision ViT as Baseline for Model Training

By

–

29 July 2023 14h03

No that I am aware (but I also haven't looked too deeply). I would probably use the original ViT in Torchvision as a baseline: https://
pytorch.org/vision/main/_m
odules/torchvision/models/vision_transformer.html#vit_b_16
… I have a companion article here showing how to accelerate the training on that model: