SambaFlow developer documentation

SambaFlow™ developer documentation includes release notes, quickstart, tutorials, and the Python API reference.

See the SambaNova Runtime documentation for information on logs, fault management, and other lower-level procedures.

Get started with SambaFlow

Release notes. Learn about new features, deprecated features, and bug fixes.
Hello SambaFlow! Compile and run a model. Learn about package installation, and how to compile and run a simple Hello World application.
Introduction to SambaFlow. Learn how SambaFlow fits into the SambaNova hardware and software stack, and about the typical compile and run workflow.
Best practices for hardware transition. If you’re migrating to SN30 hardware, the best practices help you understand changes to your models you might want to make.
SambaNova Pytorch operator support. Reference to supported Pytorch operators.

Convert existing models to SambaFlow. Learn how to convert existing models to run in your environment by exploring conversion of a Pytorch model.
Learn about model creation. Examine the Python code for a simple vision example model example.
Run example applications. Learn about SambaFlow examples and the commands to run them.
Run language example applications. Learn about language model examples and the commands to run them.

SambaNova messages and logs. Understand which messages to stdout are useful and where different logs are kept.
Use LayerNorm instead of BatchNorm. Learn how to convert a model from BatchNorm to LayerNorm.
Data parallel applications in SambaFlow. Learn about compiling a data parallel model and best practices for data parallel.
Control what tokens are back-propagated on during training. Learn how to use token ids to customize which tokens a model learns to generate, and which tokens a model attends to, but does not learn to generate.
Use Python virtual environments. Learn how to use the virtual environments that are included with SambaFlow example applications.
API Reference . SambaFlow API reference.
Data preparation scripts. We have a public GitHub repository with two scripts for pretraining data creation, pipeline.py and data_prep.py.