Topic: qwen35moe

The ggml/llama.cpp project released version b11026, which includes a model update to skip gate_up_exps if TENSOR_SKIP is set. This update is crucial for qwen35moe when MTP tensors are fused but not loaded, ensuring compatibility and functionality across different configurations.