bigmac

We have hosted the application bigmac in order to run this application in our online workstations with Wine or directly.

Quick description about bigmac:

BigMac is an open-source toolkit for pipeline-parallel training of multimodal large language models. It preserves optimized language-model pipeline schedules while placing encoder and generator work around them. This design reduces activation memory without bringing back cross-module pipeline bubbles. Its scheduler creates global operator plans, while its executor runs those plans through a shared schedule abstraction. A Megatron-Core reference backend and Qwen3 and Qwen3-VL tutorials help developers connect the approach to real training workflows. The simulator lets researchers visualize schedules, compare pipeline strategies, and model timing imbalances caused by compute cost, input size, or uneven stage partitioning. Profiling tools also expose per-operator traces for diagnosing pipeline performance before or during experiments.

Features:

• BigMac-style pipeline scheduling
• Global operator plan generation
• Megatron-Core reference execution
• Multimodal Qwen3-VL integration
• Pipeline schedule simulation
• Per-operator performance profiling

Programming Language: Python.
Categories: