0.06.101.683 I load_tensors: offloaded 47/47 layers to GPU
0.06.101.689 I load_tensors:        CUDA0 model buffer size = 56778.75 MiB
0.06.101.690 I load_tensors:        CUDA1 model buffer size = 55333.42 MiB
0.12.676.715 I llama_context:  CUDA_Host  output buffer size =     0.59 MiB
0.12.682.057 I llama_kv_cache:      CUDA0 KV buffer size =  1536.00 MiB
0.12.683.208 I llama_kv_cache:      CUDA1 KV buffer size =  1280.00 MiB
0.12.684.197 I llama_memory_recurrent:      CUDA0 RS buffer size =    77.06 MiB
0.12.684.330 I llama_memory_recurrent:      CUDA1 RS buffer size =    68.50 MiB
0.12.689.938 I llama_kv_cache:      CUDA0 KV buffer size =  1152.00 MiB
0.12.690.809 I llama_kv_cache:      CUDA1 KV buffer size =   960.00 MiB
0.12.757.516 I sched_reserve:      CUDA0 compute buffer size =   611.15 MiB
0.12.757.521 I sched_reserve:      CUDA1 compute buffer size =   611.10 MiB
0.12.757.522 I sched_reserve:  CUDA_Host compute buffer size =    86.77 MiB
