Learning has no ending
Building LLMs, diffusion models, and VLMs from scratch in PyTorch.
MoE · FP8/Triton · FSDP · DDPM/DDIM · GAN/VAE. No black boxes.
Pinned Loading
-
DeepSeek-V3-Lite
DeepSeek-V3-Lite PublicDeepSeek-V3 architecture from scratch — MLA (KV-cache compression), DeepSeekMoE (aux-loss-free balancing), FP8 Triton kernels, Multi-Token Prediction, full post-training pipeline (SFT, GRPO, R1 dis…
Python
-
mycv
mycv PublicPersonal ML Engineer portfolio — DeepSeek-V3-Lite, Stable Diffusion, VLM, GANs, all from scratch
HTML
Something went wrong, please refresh the page to try again.
If the problem persists, check the GitHub status page or contact support.
If the problem persists, check the GitHub status page or contact support.