GitHub Trending repo Repositories Jun 5 1 views

zengxiao-he/tessera: From teacher to tiles — a from-scratch LLM distillation & serving engine: custom Triton/CUDA kernels, FSDP distillation, paged-KV continuous batching, speculative decoding, a Rust gateway, a JAX oracle, and interpretability tooling.

by zengxiao-he

From teacher to tiles — a from-scratch LLM distillation & serving engine: custom Triton/CUDA kernels, FSDP distillation, paged-KV continuous batching, speculative decoding, a Rust gateway, a JAX oracle, and interpretability tooling.

Read Original

LLM NVIDIA

Metadata

Stars: 92
Forks: 1
Language: Python
Watchers: 92
License: NOASSERTION

GitHub Trending repo 8h ago

watadarkstar/react-native-nsfw-detector: A fast on device AI image safety detector for React Native / Expo using a CoreML model to detect nudity and unsafe visual content in images.

A fast on device AI image safety detector for React Native / Expo using a CoreML model to detect nudity and unsafe visual content in images.

GitHub Trending repo 19h ago

eli-labz/Third-Eye: A production-grade OSINT platform that provides situational awareness across multiple intelligence domains.

A production-grade OSINT platform that provides situational awareness across multiple intelligence domains.

GitHub Trending repo 21h ago

basinlocustmattock/dota2-predictor-analytics: Machine learning tool for predicting Dota 2 match outcomes, suggesting last picks, mining match data, training models, and visualizing hero statistics such as synergies, counters, and pick patterns.

Machine learning tool for predicting Dota 2 match outcomes, suggesting last picks, mining match data, training models, and visualizing hero statistics such as synergies, counters, ...

zengxiao-he/tessera: From teacher to tiles — a from-scratch LLM distillation & serving engine: custom Triton/CUDA kernels, FSDP distillation, paged-KV continuous batching, speculative decoding, a Rust gateway, a JAX oracle, and interpretability tooling.

Metadata

Related

watadarkstar/react-native-nsfw-detector: A fast on device AI image safety detector for React Native / Expo using a CoreML model to detect nudity and unsafe visual content in images.

eli-labz/Third-Eye: A production-grade OSINT platform that provides situational awareness across multiple intelligence domains.

basinlocustmattock/dota2-predictor-analytics: Machine learning tool for predicting Dota 2 match outcomes, suggesting last picks, mining match data, training models, and visualizing hero statistics such as synergies, counters, and pick patterns.