Heyang Qin
Source-listed: Member of Technical Staff · Anthropic
Record snapshot: Sep 3, 2026. Coverage and affiliations may be incomplete or historical.
Selected work
5- ZeRO++: Extremely Efficient Collective Communication for Giant Model TrainingSaved credit: First author · 2023Explore people connected to this work →
- DeepSpeed-FastGen: High-throughput Text Generation for LLMs via MII and DeepSpeed-InferenceSaved credit: Contributor · 2024Explore people connected to this work →
- DeepSpeed-Chat: Easy, Fast and Affordable RLHF Training of ChatGPT-like Models at All ScalesSaved credit: Contributor · 2023Explore people connected to this work →
- Phi-3 Technical Report: A Highly Capable Language Model Locally on Your PhoneSaved credit: Contributor · 2024Explore people connected to this work →
- SimiGrad: Fine-Grained Adaptive Batching for Large Scale Training using Gradient Similarity MeasurementSaved credit: First author · 2021Explore people connected to this work →