[2026-10-16 20:33:02] Load weight end. elapsed=2.54 s, type=Qwen3ForCausalLM, quant=fp8, fmt=e4m3, avail mem=13.06 GB, mem usage=8.81 GB. [2026-10-16 20:33:05] KV Cache is allocated. dtype: torch.float8_e4m3fn, #tokens: 142369, K size: 4.89 GB, V size: 4.89 GB