inference-optimization/GLM-5.3-Flash-MEP50

1---2base_model: zai-org/GLM-5.3-Flash3library_name: compressed-tensors4tags:5- moe6- expert-pruning7- compressed-tensors8---9 10# GLM-5.3-Flash - 50% Expert Pruned11 1250% of the MoE experts pruned by router-weight magnitude using [compressed-tensors](https://github.com/vllm-project/compressed-tensors).13 14- **Base model:** [zai-org/GLM-5.3-Flash](https://huggingface.co/zai-org/GLM-5.3-Flash)15- **Sparsity:** 50% of routed experts removed per layer16- **Layers pruned:** all 43 MoE layers (body layers 3-44)17- **MTP layer:** retained exactly as-is18- **Shared experts:** untouched19- **Vision tower:** untouched (dense ViT MLP, no MoE experts)20 21## Reproduction22 23 24