D-Matrix Corsair: delivering low latency batched inference for inference-time-compute

GPU MODE - D-Matrix Corsair: delivering low latency batched inference for inference-time-compute