`bridge.recipes.deepseek.deepseek_v4`#

Module Contents#

Functions#

`set_deepseek_v4_pipeline_model_parallel_layout`	Set an even DSv4 pipeline layout with MTP and loss on the last stage.
`_deepseek_v4_mxfp8_quant_recipe`	Use MXFP8 for training and BF16 for DSv4 validation/evaluation paths.
`deepseek_v4_flash_pretrain_config`	Return the DeepSeek-V4-Flash Blackwell pre-training base config.
`deepseek_v4_flash_pretrain_mxfp8_config`	Return the DeepSeek-V4-Flash Adam + MXFP8 pre-training config.
`deepseek_v4_flash_pretrain_muon_config`	Return the DeepSeek-V4-Flash BF16 Muon pre-training config.

API#

bridge.recipes.deepseek.deepseek_v4.set_deepseek_v4_pipeline_model_parallel_layout( model_cfg: megatron.bridge.models.GPTModelProvider, ) → None#: Set an even DSv4 pipeline layout with MTP and loss on the last stage.

bridge.recipes.deepseek.deepseek_v4._deepseek_v4_mxfp8_quant_recipe() → megatron.core.quantization.quant_config.RecipeConfig#: Use MXFP8 for training and BF16 for DSv4 validation/evaluation paths.

bridge.recipes.deepseek.deepseek_v4.deepseek_v4_flash_pretrain_config() → megatron.bridge.training.config.ConfigContainer#

Return the DeepSeek-V4-Flash Blackwell pre-training base config.

Recommended Blackwell baseline: TP=1, PP=4, EP=8, CP=1.

bridge.recipes.deepseek.deepseek_v4.deepseek_v4_flash_pretrain_mxfp8_config() → megatron.bridge.training.config.ConfigContainer#: Return the DeepSeek-V4-Flash Adam + MXFP8 pre-training config.

bridge.recipes.deepseek.deepseek_v4.deepseek_v4_flash_pretrain_muon_config() → megatron.bridge.training.config.ConfigContainer#: Return the DeepSeek-V4-Flash BF16 Muon pre-training config.