LLMpeople
Public Atlas People first, reports as evidence, organizations as context.

Atlas / People / Detail

Yanghua Peng

Public sources list Yanghua Peng as a coauthor on ByteDance Seed technical reports and on ByteDance engineering research artifacts related to distributed training and checkpointing for large models.

ByteDance coauthor on large-model training infrastructure and Seed technical reports.1 organizations2 reports

Profile status: updated

updated 4 public sources
multimodal modelsreasoning modelsdistributed training systemscheckpointing systems

Current frame

ByteDance coauthor on large-model training infrastructure and Seed technical reports.

Extended note

Yanghua Peng is publicly listed as a coauthor of the ByteDance Seed reports "Seed1.5-VL Technical Report" and "Seed1.5-Thinking: Advancing Superb Reasoning Models with Reinforcement Learning." Public ByteDance GitHub repositories also attribute Yanghua Peng on work tied to distributed training and checkpointing systems, including BytePS and ByteCheckpoint. Because no directly attributable personal profile was found within the exact-source budget, this profile keeps biography claims narrow and leaves personal links, education, and work history unset.

Organizations

core ByteDance Seed

Reports

Vision-Language Models Seed1.5-VL Technical Report Report Seed1.5-Thinking: Advancing Superb Reasoning Models with Reinforcement Learning

Supporting sources

ByteDance-Seed/ByteCheckpoint: ByteCheckpoint: A Unified Checkpointing System for Large Foundation Model Development Supporting source · github · GitHub bytedance/byteps: A high performance and generic framework for distributed DNN training Supporting source · github · GitHub Seed1.5-Thinking: Advancing Superb Reasoning Models with Reinforcement Learning Supporting source · report · arXiv Seed1.5-VL Technical Report Supporting source · report · arXiv