[2026-10-16 19:26:38] Load weight end. elapsed=3.92 s, type=Qwen3ForCausalLM, avail mem=6.59 GB, mem usage=15.28 GB. [2026-10-16 19:26:40] KV Cache is allocated. dtype: torch.bfloat16, #tokens: 24094, K size: 1.65 GB, V size: 1.65 GB