CookLLM
Docs
Pricing
Roadmap
Commits
About Me
Toggle mode
Switch language
Roadmap
Explore our future plans and upcoming course modules
Overall progress
10/26 completed
38%
Backlog
9
Modern Transformer tricks doc
low
Harry
KV Cache & text generation doc
medium
Harry
Evaluation metrics doc
low
Harry
RLHF: PPO training doc and code
medium
Harry
DPO (Direct Preference Optimization) doc and code
medium
Harry
GRPO (Group Relative Policy Optimization) doc and code
medium
Harry
Build a mini vLLM inference engine
high
Harry
Multimodal model doc and code
medium
Harry
LLM Agent framework doc and code
medium
Harry
In Progress
7
Introduce training parallelism module
high
Harry
2026-02-01
Introduce modern LLM model architectures designs
high
Harry
2026-02-01
Normalization doc (BatchNorm → RMSNorm)
medium
Harry
Activation functions doc (ReLU → SwiGLU)
medium
Harry
Optimizer theory doc (Adam → Muon)
medium
Harry
Training techniques doc (LR schedule, mixed precision)
medium
Harry
SFT (Supervised Fine-Tuning) doc and code
medium
Harry
Done
10
Introduce LLM tokenizer
high
Harry
2026-01-17
Training tokenizer from scratch
high
Harry
2026-01-17
Flash Attention doc and code
high
Harry
2025-12-15
GPU programming doc and code
high
Harry
2025-12-15
RoPE position encoding doc and code
high
Harry
2026-02-10
build the cookllm-bento codebase
high
Harry
2026-01-28
Training: tokenizer hands-on (RustBPE + tiktoken)
high
Harry
2026-02-05
Training: BentoLM architecture design & implementation
high
Harry
2026-02-14
Training: pretrain pipeline (data, loop, monitoring)
high
Harry
2026-02-20
Training: pretrain inference & text generation
medium
Harry
2026-02-22