bridge.models.glm_moe_dsa.glm5_bridge

Module Contents

Classes

GLM5Bridge

Megatron Bridge for GLM-5 (MoE + MLA + DSA).

Data

logger

API

bridge.models.glm_moe_dsa.glm5_bridge.logger

‘getLogger(…)’

class bridge.models.glm_moe_dsa.glm5_bridge.GLM5Bridge

Bases: megatron.bridge.models.conversion.model_bridge.MegatronModelBridge

Megatron Bridge for GLM-5 (MoE + MLA + DSA).

This bridge handles conversion between HuggingFace GlmMoeDsaForCausalLM and Megatron-Core GPTModel formats.

GLM-5 uses Multi-Latent Attention (MLA), Dynamic Sparse Attention (DSA) indexer layers, and Mixture-of-Experts (MoE). Requires transformers>=5.2.0.

.. rubric:: Example

from megatron.bridge import AutoBridge bridge = AutoBridge.from_hf_pretrained(“zai-org/GLM-5”) provider = bridge.to_megatron_provider()

provider_bridge(

hf_pretrained: megatron.bridge.models.hf_pretrained.causal_lm.PreTrainedCausalLM,

) → megatron.bridge.models.mla_provider.MLAModelProvider

mapping_registry() → megatron.bridge.models.conversion.mapping_registry.MegatronMappingRegistry