bitnet triton
Custom Triton kernel for 1.58-bit LLM inference — 4.4× less VRAM, 1.5× faster decode, +0.26% PPL on RTX 4060 Laptop
Python
🔥
140
Builder Score
⭐
4
GitHub Stars
🏆
0
Hackathons Won
🔀
0
Open Source PRs
💻
7k
Lines of Code
▲
0
Total Upvotes
Custom Triton kernel for 1.58-bit LLM inference — 4.4× less VRAM, 1.5× faster decode, +0.26% PPL on RTX 4060 Laptop
Cross-platform Material-Design chat app with a threaded Python socket server, SQLite persistence, and PBKDF2-hashed acco…
Custom CNN + Bidirectional LSTM speech recognition model in TensorFlow, packaged for distributed training on Google Clou…