bridge.recipes.glm.h100.glm45

Module Contents

Functions

glm45_355b_pretrain_128gpu_h100_bf16_config	Return a pre-training config for GLM 4.5 355B-A32B variant.
glm45_air_106b_pretrain_32gpu_h100_bf16_config	Return a pre-training config for GLM 4.5 Air 106B-A12B variant.
glm45_355b_sft_128gpu_h100_bf16_config	Return a full SFT config for GLM-4.5 355B-A32B.
glm45_air_106b_sft_32gpu_h100_bf16_config	Return a full SFT config for GLM-4.5 Air 106B-A12B.
glm45_355b_peft_16gpu_h100_bf16_config	Return a PEFT config for GLM-4.5 355B-A32B.
glm45_air_106b_peft_8gpu_h100_bf16_config	Return a PEFT config for GLM-4.5 Air 106B-A12B.

Data

__all__

API

bridge.recipes.glm.h100.glm45.glm45_355b_pretrain_128gpu_h100_bf16_config() → megatron.bridge.training.config.ConfigContainer

Return a pre-training config for GLM 4.5 355B-A32B variant.

Recommended parallelism: TP=2, PP=8, EP=16

bridge.recipes.glm.h100.glm45.glm45_air_106b_pretrain_32gpu_h100_bf16_config() → megatron.bridge.training.config.ConfigContainer

Return a pre-training config for GLM 4.5 Air 106B-A12B variant.

Recommended parallelism: TP=1, PP=4, EP=8

bridge.recipes.glm.h100.glm45.glm45_355b_sft_128gpu_h100_bf16_config() → megatron.bridge.training.config.ConfigContainer

Return a full SFT config for GLM-4.5 355B-A32B.

Default parallelism: TP=2, PP=8, EP=16

Note: Packed sequence is NOT supported for GLM-4.5.

Returns:

ConfigContainer with all settings pre-configured for GLM-4.5 355B SFT.

bridge.recipes.glm.h100.glm45.glm45_air_106b_sft_32gpu_h100_bf16_config() → megatron.bridge.training.config.ConfigContainer

Return a full SFT config for GLM-4.5 Air 106B-A12B.

Default parallelism: TP=1, PP=4, EP=8

Note: Packed sequence is NOT supported for GLM-4.5.

Returns:

ConfigContainer with all settings pre-configured for GLM-4.5 Air 106B SFT.

bridge.recipes.glm.h100.glm45.glm45_355b_peft_16gpu_h100_bf16_config(

peft_scheme: str | megatron.bridge.peft.base.PEFT = 'lora',

) → megatron.bridge.training.config.ConfigContainer

Return a PEFT config for GLM-4.5 355B-A32B.

Default parallelism: TP=2, PP=4, EP=4

Note: Packed sequence is NOT supported for GLM-4.5.

Parameters:

peft_scheme – PEFT scheme - “lora”, “dora”, or a custom PEFT instance.

Returns:

ConfigContainer with all settings pre-configured for GLM-4.5 355B PEFT.

bridge.recipes.glm.h100.glm45.glm45_air_106b_peft_8gpu_h100_bf16_config(

peft_scheme: str | megatron.bridge.peft.base.PEFT = 'lora',

) → megatron.bridge.training.config.ConfigContainer

Return a PEFT config for GLM-4.5 Air 106B-A12B.

Default parallelism: TP=1, PP=2, EP=4

Note: Packed sequence is NOT supported for GLM-4.5.

Parameters:

peft_scheme – PEFT scheme - “lora”, “dora”, or a custom PEFT instance.

Returns:

ConfigContainer with all settings pre-configured for GLM-4.5 Air 106B PEFT.

bridge.recipes.glm.h100.glm45.__all__

[‘glm45_355b_peft_16gpu_h100_bf16_config’, ‘glm45_355b_pretrain_128gpu_h100_bf16_config’, ‘glm45_355…