`bridge.perf_recipes.llama.common`#

Common helpers for llama performance recipes.

Module Contents#

`_with_global_batch_size`
`_llama_benchmark_common`	Apply legacy Llama benchmark defaults shared by the flat recipes.

bridge.perf_recipes.llama.common._with_global_batch_size( cfg: megatron.bridge.training.config.ConfigContainer, global_batch_size: int, ) → megatron.bridge.training.config.ConfigContainer#

bridge.perf_recipes.llama.common._llama_benchmark_common( cfg: megatron.bridge.training.config.ConfigContainer, ) → None#: Apply legacy Llama benchmark defaults shared by the flat recipes.