Implement a ChatGPT-like LLM in PyTorch from scratch, step by step
COMMITS
May 23, 2026
T
Add Deepseek Sparse Attention (DSA) implementation (ch04/09_dsa) (#1014)
Tarun kumar committed
May 19, 2026
S
Muon optimizer (#1029)
Sebastian Raschka committed
S
Fix TensorFlow dependency marker for Linux ARM (#1028)
Sebastian Raschka committed
May 17, 2026
S
Tips on forking and updating this repo locally (#1027)
Sebastian Raschka committed
S
Unique names to improve automerge if all tests pass (#1026)
Sebastian Raschka committed
S
Add cross-layer KV sharing (#1025)
Sebastian Raschka committed
May 11, 2026
S
Add troubleshooting guide for common questions (#1024)
Sebastian Raschka committed
May 6, 2026
A
Fix typo in README for KV-cache totals (#1023)
Anthony Chan committed
April 16, 2026
S
Fix SWA (#1013)
Sebastian Raschka committed
April 11, 2026
A
Fix BPE empty-pair edge case in find_freq_pair (#1004)
Amit Choubey committed
S
Gemma 4 (#1005)
Sebastian Raschka committed
April 6, 2026
S
Some gemma 3 improvements (#1000)
Sebastian Raschka committed
April 4, 2026
S
Remove 7 days requirement to improve windows compatibility (#999)
Sebastian Raschka committed
April 1, 2026
S
uv packages must be 7 days old (#994)
Sebastian Raschka committed
March 31, 2026
S
Swap urllib.request with requests (#993)
Sebastian Raschka committed
March 26, 2026
D
fix: pin 1 unpinned action(s) (#987)
dagecko committed
March 21, 2026
C
fix: added KVcache in `generate_text_basic_stream` (#981)
casinca committed
March 7, 2026
R
harded the link checker
rasbt committed
S
Minor typo fix (#974)
Sebastian Raschka committed
S
Bpe whitespace fixes (#975)
Sebastian Raschka committed
March 4, 2026
S
Add more analysis to qwen3.5 image
Sebastian Raschka committed
March 3, 2026
S
Use full HF url
Sebastian Raschka committed
S
Qwen3.5 from scratch (#969)
Sebastian Raschka committed
March 2, 2026
R
User argpars utils to show default args on command line
rasbt committed
February 27, 2026
S
Jupyter scrolling glitch tips (#965)
Sebastian Raschka committed
February 19, 2026
R
image size
rasbt committed
R
image size
rasbt committed
R
formatting fix
rasbt committed
S
Add Tiny Aya from scratch (#962)
Sebastian Raschka committed
S
remove redundant assignment (#961)
Sebastian Raschka committed