Does anyone else have their models get stuck in loops like this? I was trying to bake off a 3080 Ti(CUDA13) with Qwen3.5-9B vs and a Xe iGPU with Qwen3.5-35B-A3B.
