Implement a ChatGPT-like LLM in PyTorch from scratch, step by step
COMMITS
February 12, 2025
S
PyTorch tips for better training performance (#525)
Sebastian Raschka committed
February 11, 2025
S
Add torchrun bonus code (#524)
Sebastian Raschka committed
V
Comment that DDP-script.py does not work with GPUs > 2 (#523)
Victor Skvortsov committed
February 9, 2025
S
Upgrade to NumPy 2.0 (#520)
Sebastian Raschka committed
February 6, 2025
S
Mention small discrepancy due to Dropout non-reproducibility in PyTorch (#519)
Sebastian Raschka committed
D
fixed indention and enumeration for nvct (#518)
Daniel Kleine committed
February 5, 2025
S
Fix typo in Ch02 comments (#516)
Sebastian Raschka committed
February 1, 2025
S
More pythonic way to find the longest sequence (#512)
Sebastian Raschka committed
January 30, 2025
S
Test PyTorch nightly releases (#509)
Sebastian Raschka committed
January 29, 2025
S
Alternative weight loading via .safetensors (#507)
Sebastian Raschka committed
January 25, 2025
S
Fix default argument in ex 7.2 (#506)
Sebastian Raschka committed
January 23, 2025
S
A few cosmetic updates (#504)
Sebastian Raschka committed
S
Test for PyTorch 2.6 release candidate (#500)
Sebastian Raschka committed
January 22, 2025
S
Bonus material: extending tokenizers (#496)
Sebastian Raschka committed
D
add GPT2TokenizerFast to BPE comparison (#498)
Daniel Kleine committed
January 21, 2025
A
fix: preserve newline tokens in BPE encoder (#495)
Austin Welch committed
D
BPE: fixed typo (#492)
Daniel Kleine committed
January 17, 2025
S
Implementingthe BPE Tokenizer from Scratch (#487)
Sebastian Raschka committed
R
04_optional-aws-sagemaker-notebook (#451)
rvaneijk committed
January 15, 2025
S
Include mathematical breakdown for exercise solution 4.1 (#483)
Sebastian Raschka committed
January 13, 2025
H
Print out embeddings for more illustrative learning (#481)
Henry Shi committed
January 12, 2025
R
fix reward margins plot label in dpo nb
rasbt committed
S
Auto download DPO dataset if not already available in path (#479)
Sebastian Raschka committed
January 8, 2025
S
adds no-grad context for reference model to DPO (#473)
Sebastian Raschka committed
January 5, 2025
S
fix ch07 unit test (#470)
Sebastian Raschka committed
S
Add backup URL for gpt2 weights (#469)
Sebastian Raschka committed
Q
typo fixed (#468)
QS committed
January 4, 2025
T
Minor readability improvement in dataloader.ipynb (#461)
Tao Qian committed
S
fix misplaced parenthesis and update license (#466)
Sebastian Raschka committed
November 18, 2024
C
[minor] typo & comments (#441)
casinca committed