| INFO 07-07 00:47:11 [__init__.py:239] Automatically detected platform cuda. | |
| INFO 07-07 00:47:13 [config.py:209] Replacing legacy 'type' key with 'rope_type' | |
| INFO 07-07 00:47:13 [config.py:2968] Downcasting torch.float32 to torch.float16. | |
| INFO 07-07 00:47:20 [config.py:717] This model supports multiple tasks: {'classify', 'reward', 'embed', 'generate', 'score'}. Defaulting to 'generate'. | |
| INFO 07-07 00:47:20 [config.py:1770] Defaulting to use mp for distributed inference | |
| INFO 07-07 00:47:20 [config.py:2003] Chunked prefill is enabled with max_num_batched_tokens=16384. | |
| INFO 07-07 00:47:22 [core.py:58] Initializing a V1 LLM engine (v0.8.5.post1) with config: model='./merged1/phi_linear_7', speculative_config=None, tokenizer='./merged1/phi_linear_7', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, override_neuron_config=None, tokenizer_revision=None, trust_remote_code=False, dtype=torch.float16, max_seq_len=2048, download_dir=None, load_format=auto, tensor_parallel_size=4, pipeline_parallel_size=1, disable_custom_all_reduce=False, quantization=None, enforce_eager=False, kv_cache_dtype=auto, device_config=cuda, decoding_config=DecodingConfig(guided_decoding_backend='auto', reasoning_backend=None), observability_config=ObservabilityConfig(show_hidden_metrics=False, otlp_traces_endpoint=None, collect_model_forward_time=False, collect_model_execute_time=False), seed=None, served_model_name=./merged1/phi_linear_7, num_scheduler_steps=1, multi_step_stream_outputs=True, enable_prefix_caching=True, chunked_prefill_enabled=True, use_async_output_proc=True, disable_mm_preprocessor_cache=False, mm_processor_kwargs=None, pooler_config=None, compilation_config={"level":3,"custom_ops":["none"],"splitting_ops":["vllm.unified_attention","vllm.unified_attention_with_output"],"use_inductor":true,"compile_sizes":[],"use_cudagraph":true,"cudagraph_num_of_warmups":1,"cudagraph_capture_sizes":[512,504,496,488,480,472,464,456,448,440,432,424,416,408,400,392,384,376,368,360,352,344,336,328,320,312,304,296,288,280,272,264,256,248,240,232,224,216,208,200,192,184,176,168,160,152,144,136,128,120,112,104,96,88,80,72,64,56,48,40,32,24,16,8,4,2,1],"max_capture_size":512} | |
| WARNING 07-07 00:47:22 [multiproc_worker_utils.py:306] Reducing Torch parallelism from 128 threads to 1 to avoid unnecessary CPU contention. Set OMP_NUM_THREADS in the external environment to tune this value as needed. | |
| INFO 07-07 00:47:22 [shm_broadcast.py:266] vLLM message queue communication handle: Handle(local_reader_ranks=[0, 1, 2, 3], buffer_handle=(4, 10485760, 10, 'psm_8566a3d6'), local_subscribe_addr='ipc:///tmp/1273a10f-2388-49d5-b4c3-3f7d4b8dc539', remote_subscribe_addr=None, remote_addr_ipv6=False) | |
| WARNING 07-07 00:47:22 [utils.py:2522] Methods determine_num_available_blocks,device_config,get_cache_block_size_bytes,initialize_cache not implemented in <vllm.v1.worker.gpu_worker.Worker object at 0x154c023affd0> | |
| [1;36m(VllmWorker rank=1 pid=4062166)[0;0m INFO 07-07 00:47:22 [shm_broadcast.py:266] vLLM message queue communication handle: Handle(local_reader_ranks=[0], buffer_handle=(1, 10485760, 10, 'psm_4db68858'), local_subscribe_addr='ipc:///tmp/b41ed844-377e-49fa-9b53-77b2f2cdc8ec', remote_subscribe_addr=None, remote_addr_ipv6=False) | |
| WARNING 07-07 00:47:22 [utils.py:2522] Methods determine_num_available_blocks,device_config,get_cache_block_size_bytes,initialize_cache not implemented in <vllm.v1.worker.gpu_worker.Worker object at 0x154c00984d90> | |
| [1;36m(VllmWorker rank=0 pid=4062165)[0;0m INFO 07-07 00:47:22 [shm_broadcast.py:266] vLLM message queue communication handle: Handle(local_reader_ranks=[0], buffer_handle=(1, 10485760, 10, 'psm_b61e5e10'), local_subscribe_addr='ipc:///tmp/3821bfbc-7c30-4a0d-af43-8fc058da5e24', remote_subscribe_addr=None, remote_addr_ipv6=False) | |
| WARNING 07-07 00:47:22 [utils.py:2522] Methods determine_num_available_blocks,device_config,get_cache_block_size_bytes,initialize_cache not implemented in <vllm.v1.worker.gpu_worker.Worker object at 0x154c023aff70> | |
| WARNING 07-07 00:47:22 [utils.py:2522] Methods determine_num_available_blocks,device_config,get_cache_block_size_bytes,initialize_cache not implemented in <vllm.v1.worker.gpu_worker.Worker object at 0x154c023afd30> | |
| [1;36m(VllmWorker rank=2 pid=4062167)[0;0m INFO 07-07 00:47:22 [shm_broadcast.py:266] vLLM message queue communication handle: Handle(local_reader_ranks=[0], buffer_handle=(1, 10485760, 10, 'psm_e108a397'), local_subscribe_addr='ipc:///tmp/e970d196-5ca3-4ce0-bb7e-bf339d4ccc25', remote_subscribe_addr=None, remote_addr_ipv6=False) | |
| [1;36m(VllmWorker rank=3 pid=4062168)[0;0m INFO 07-07 00:47:22 [shm_broadcast.py:266] vLLM message queue communication handle: Handle(local_reader_ranks=[0], buffer_handle=(1, 10485760, 10, 'psm_34c134a9'), local_subscribe_addr='ipc:///tmp/665b2bc9-4d0b-4c99-9a76-fb1479cd7315', remote_subscribe_addr=None, remote_addr_ipv6=False) | |
| [1;36m(VllmWorker rank=1 pid=4062166)[0;0m INFO 07-07 00:47:25 [utils.py:1055] Found nccl from library libnccl.so.2 | |
| [1;36m(VllmWorker rank=0 pid=4062165)[0;0m INFO 07-07 00:47:25 [utils.py:1055] Found nccl from library libnccl.so.2 | |
| [1;36m(VllmWorker rank=1 pid=4062166)[0;0m INFO 07-07 00:47:25 [pynccl.py:69] vLLM is using nccl==2.21.5 | |
| [1;36m(VllmWorker rank=0 pid=4062165)[0;0m INFO 07-07 00:47:25 [pynccl.py:69] vLLM is using nccl==2.21.5 | |
| [1;36m(VllmWorker rank=3 pid=4062168)[0;0m INFO 07-07 00:47:25 [utils.py:1055] Found nccl from library libnccl.so.2 | |
| [1;36m(VllmWorker rank=2 pid=4062167)[0;0m INFO 07-07 00:47:25 [utils.py:1055] Found nccl from library libnccl.so.2 | |
| [1;36m(VllmWorker rank=3 pid=4062168)[0;0m INFO 07-07 00:47:25 [pynccl.py:69] vLLM is using nccl==2.21.5 | |
| [1;36m(VllmWorker rank=2 pid=4062167)[0;0m INFO 07-07 00:47:25 [pynccl.py:69] vLLM is using nccl==2.21.5 | |
| [1;36m(VllmWorker rank=2 pid=4062167)[0;0m WARNING 07-07 00:47:25 [custom_all_reduce.py:136] Custom allreduce is disabled because it's not supported on more than two PCIe-only GPUs. To silence this warning, specify disable_custom_all_reduce=True explicitly. | |
| [1;36m(VllmWorker rank=3 pid=4062168)[0;0m WARNING 07-07 00:47:25 [custom_all_reduce.py:136] Custom allreduce is disabled because it's not supported on more than two PCIe-only GPUs. To silence this warning, specify disable_custom_all_reduce=True explicitly. | |
| [1;36m(VllmWorker rank=1 pid=4062166)[0;0m WARNING 07-07 00:47:25 [custom_all_reduce.py:136] Custom allreduce is disabled because it's not supported on more than two PCIe-only GPUs. To silence this warning, specify disable_custom_all_reduce=True explicitly. | |
| [1;36m(VllmWorker rank=0 pid=4062165)[0;0m WARNING 07-07 00:47:25 [custom_all_reduce.py:136] Custom allreduce is disabled because it's not supported on more than two PCIe-only GPUs. To silence this warning, specify disable_custom_all_reduce=True explicitly. | |
| [1;36m(VllmWorker rank=0 pid=4062165)[0;0m INFO 07-07 00:47:25 [shm_broadcast.py:266] vLLM message queue communication handle: Handle(local_reader_ranks=[1, 2, 3], buffer_handle=(3, 4194304, 6, 'psm_959315af'), local_subscribe_addr='ipc:///tmp/697cd960-09d8-4c74-8c9e-995091c8b87d', remote_subscribe_addr=None, remote_addr_ipv6=False) | |
| [1;36m(VllmWorker rank=3 pid=4062168)[0;0m INFO 07-07 00:47:25 [parallel_state.py:1004] rank 3 in world size 4 is assigned as DP rank 0, PP rank 0, TP rank 3 | |
| [1;36m(VllmWorker rank=2 pid=4062167)[0;0m INFO 07-07 00:47:25 [parallel_state.py:1004] rank 2 in world size 4 is assigned as DP rank 0, PP rank 0, TP rank 2 | |
| [1;36m(VllmWorker rank=1 pid=4062166)[0;0m INFO 07-07 00:47:25 [parallel_state.py:1004] rank 1 in world size 4 is assigned as DP rank 0, PP rank 0, TP rank 1 | |
| [1;36m(VllmWorker rank=0 pid=4062165)[0;0m INFO 07-07 00:47:25 [parallel_state.py:1004] rank 0 in world size 4 is assigned as DP rank 0, PP rank 0, TP rank 0 | |
| [1;36m(VllmWorker rank=3 pid=4062168)[0;0m INFO 07-07 00:47:25 [cuda.py:221] Using Flash Attention backend on V1 engine. | |
| [1;36m(VllmWorker rank=2 pid=4062167)[0;0m INFO 07-07 00:47:25 [cuda.py:221] Using Flash Attention backend on V1 engine. | |
| [1;36m(VllmWorker rank=3 pid=4062168)[0;0m WARNING 07-07 00:47:25 [topk_topp_sampler.py:69] FlashInfer is not available. Falling back to the PyTorch-native implementation of top-p & top-k sampling. For the best performance, please install FlashInfer. | |
| [1;36m(VllmWorker rank=2 pid=4062167)[0;0m WARNING 07-07 00:47:25 [topk_topp_sampler.py:69] FlashInfer is not available. Falling back to the PyTorch-native implementation of top-p & top-k sampling. For the best performance, please install FlashInfer. | |
| [1;36m(VllmWorker rank=1 pid=4062166)[0;0m INFO 07-07 00:47:25 [cuda.py:221] Using Flash Attention backend on V1 engine. | |
| [1;36m(VllmWorker rank=1 pid=4062166)[0;0m WARNING 07-07 00:47:25 [topk_topp_sampler.py:69] FlashInfer is not available. Falling back to the PyTorch-native implementation of top-p & top-k sampling. For the best performance, please install FlashInfer. | |
| [1;36m(VllmWorker rank=0 pid=4062165)[0;0m INFO 07-07 00:47:25 [cuda.py:221] Using Flash Attention backend on V1 engine. | |
| [1;36m(VllmWorker rank=0 pid=4062165)[0;0m WARNING 07-07 00:47:25 [topk_topp_sampler.py:69] FlashInfer is not available. Falling back to the PyTorch-native implementation of top-p & top-k sampling. For the best performance, please install FlashInfer. | |
| [1;36m(VllmWorker rank=1 pid=4062166)[0;0m INFO 07-07 00:47:25 [gpu_model_runner.py:1329] Starting to load model ./merged1/phi_linear_7... | |
| [1;36m(VllmWorker rank=2 pid=4062167)[0;0m INFO 07-07 00:47:25 [gpu_model_runner.py:1329] Starting to load model ./merged1/phi_linear_7... | |
| [1;36m(VllmWorker rank=3 pid=4062168)[0;0m INFO 07-07 00:47:25 [gpu_model_runner.py:1329] Starting to load model ./merged1/phi_linear_7... | |
| [1;36m(VllmWorker rank=0 pid=4062165)[0;0m INFO 07-07 00:47:25 [gpu_model_runner.py:1329] Starting to load model ./merged1/phi_linear_7... | |
| [1;36m(VllmWorker rank=1 pid=4062166)[0;0m INFO 07-07 00:47:27 [loader.py:458] Loading weights took 1.64 seconds | |
| [1;36m(VllmWorker rank=0 pid=4062165)[0;0m INFO 07-07 00:47:27 [loader.py:458] Loading weights took 1.85 seconds | |
| [1;36m(VllmWorker rank=3 pid=4062168)[0;0m INFO 07-07 00:47:27 [loader.py:458] Loading weights took 1.91 seconds | |
| [1;36m(VllmWorker rank=1 pid=4062166)[0;0m INFO 07-07 00:47:28 [gpu_model_runner.py:1347] Model loading took 1.8196 GiB and 1.866224 seconds | |
| [1;36m(VllmWorker rank=2 pid=4062167)[0;0m INFO 07-07 00:47:28 [loader.py:458] Loading weights took 2.03 seconds | |
| [1;36m(VllmWorker rank=3 pid=4062168)[0;0m INFO 07-07 00:47:28 [gpu_model_runner.py:1347] Model loading took 1.8196 GiB and 2.107000 seconds | |
| [1;36m(VllmWorker rank=2 pid=4062167)[0;0m INFO 07-07 00:47:28 [gpu_model_runner.py:1347] Model loading took 1.8196 GiB and 2.223928 seconds | |
| [1;36m(VllmWorker rank=0 pid=4062165)[0;0m INFO 07-07 00:47:28 [gpu_model_runner.py:1347] Model loading took 1.8196 GiB and 2.072886 seconds | |
| [1;36m(VllmWorker rank=2 pid=4062167)[0;0m INFO 07-07 00:47:34 [backends.py:420] Using cache directory: /home/jiangli/.cache/vllm/torch_compile_cache/39e4f38180/rank_2_0 for vLLM's torch.compile | |
| [1;36m(VllmWorker rank=2 pid=4062167)[0;0m INFO 07-07 00:47:34 [backends.py:430] Dynamo bytecode transform time: 5.61 s | |
| [1;36m(VllmWorker rank=3 pid=4062168)[0;0m INFO 07-07 00:47:34 [backends.py:420] Using cache directory: /home/jiangli/.cache/vllm/torch_compile_cache/39e4f38180/rank_3_0 for vLLM's torch.compile | |
| [1;36m(VllmWorker rank=3 pid=4062168)[0;0m INFO 07-07 00:47:34 [backends.py:430] Dynamo bytecode transform time: 5.64 s | |
| [1;36m(VllmWorker rank=0 pid=4062165)[0;0m INFO 07-07 00:47:34 [backends.py:420] Using cache directory: /home/jiangli/.cache/vllm/torch_compile_cache/39e4f38180/rank_0_0 for vLLM's torch.compile | |
| [1;36m(VllmWorker rank=0 pid=4062165)[0;0m INFO 07-07 00:47:34 [backends.py:430] Dynamo bytecode transform time: 5.83 s | |
| [1;36m(VllmWorker rank=1 pid=4062166)[0;0m INFO 07-07 00:47:34 [backends.py:420] Using cache directory: /home/jiangli/.cache/vllm/torch_compile_cache/39e4f38180/rank_1_0 for vLLM's torch.compile | |
| [1;36m(VllmWorker rank=1 pid=4062166)[0;0m INFO 07-07 00:47:34 [backends.py:430] Dynamo bytecode transform time: 5.94 s | |
| [1;36m(VllmWorker rank=2 pid=4062167)[0;0m INFO 07-07 00:47:39 [backends.py:118] Directly load the compiled graph(s) for shape None from the cache, took 4.615 s | |
| [1;36m(VllmWorker rank=3 pid=4062168)[0;0m INFO 07-07 00:47:39 [backends.py:118] Directly load the compiled graph(s) for shape None from the cache, took 4.647 s | |
| [1;36m(VllmWorker rank=0 pid=4062165)[0;0m INFO 07-07 00:47:39 [backends.py:118] Directly load the compiled graph(s) for shape None from the cache, took 4.614 s | |
| [1;36m(VllmWorker rank=1 pid=4062166)[0;0m INFO 07-07 00:47:39 [backends.py:118] Directly load the compiled graph(s) for shape None from the cache, took 4.609 s | |
| [1;36m(VllmWorker rank=3 pid=4062168)[0;0m INFO 07-07 00:47:45 [monitor.py:33] torch.compile takes 5.64 s in total | |
| [1;36m(VllmWorker rank=2 pid=4062167)[0;0m INFO 07-07 00:47:45 [monitor.py:33] torch.compile takes 5.61 s in total | |
| [1;36m(VllmWorker rank=1 pid=4062166)[0;0m INFO 07-07 00:47:45 [monitor.py:33] torch.compile takes 5.94 s in total | |
| [1;36m(VllmWorker rank=0 pid=4062165)[0;0m INFO 07-07 00:47:45 [monitor.py:33] torch.compile takes 5.83 s in total | |
| INFO 07-07 00:47:46 [kv_cache_utils.py:634] GPU KV cache size: 2,007,088 tokens | |
| INFO 07-07 00:47:46 [kv_cache_utils.py:637] Maximum concurrency for 2,048 tokens per request: 980.02x | |
| INFO 07-07 00:47:46 [kv_cache_utils.py:634] GPU KV cache size: 2,006,832 tokens | |
| INFO 07-07 00:47:46 [kv_cache_utils.py:637] Maximum concurrency for 2,048 tokens per request: 979.90x | |
| INFO 07-07 00:47:46 [kv_cache_utils.py:634] GPU KV cache size: 2,006,832 tokens | |
| INFO 07-07 00:47:46 [kv_cache_utils.py:637] Maximum concurrency for 2,048 tokens per request: 979.90x | |
| INFO 07-07 00:47:46 [kv_cache_utils.py:634] GPU KV cache size: 2,008,112 tokens | |
| INFO 07-07 00:47:46 [kv_cache_utils.py:637] Maximum concurrency for 2,048 tokens per request: 980.52x | |
| [1;36m(VllmWorker rank=2 pid=4062167)[0;0m INFO 07-07 00:48:13 [gpu_model_runner.py:1686] Graph capturing finished in 26 secs, took 3.00 GiB | |
| [1;36m(VllmWorker rank=3 pid=4062168)[0;0m INFO 07-07 00:48:13 [gpu_model_runner.py:1686] Graph capturing finished in 26 secs, took 3.00 GiB | |
| [1;36m(VllmWorker rank=0 pid=4062165)[0;0m INFO 07-07 00:48:13 [gpu_model_runner.py:1686] Graph capturing finished in 26 secs, took 3.00 GiB | |
| [1;36m(VllmWorker rank=1 pid=4062166)[0;0m INFO 07-07 00:48:13 [gpu_model_runner.py:1686] Graph capturing finished in 27 secs, took 3.00 GiB | |
| INFO 07-07 00:48:13 [core.py:159] init engine (profile, create kv cache, warmup model) took 45.04 seconds | |
| INFO 07-07 00:48:13 [core_client.py:439] Core engine process 0 ready. | |
| INFO 07-07 00:49:48 [importing.py:53] Triton module has been replaced with a placeholder. | |
| INFO 07-07 00:49:49 [__init__.py:239] Automatically detected platform cuda. | |
| | Task |Version| Metric |Value | |Stderr| | |
| |------------------|------:|---------------------|-----:|---|-----:| | |
| |all | |sem |0.8347|± |0.0219| | |
| | | |math_pass@1:1_samples|0.9372|± |0.0196| | |
| |mm\|arc_challenge\|0| 0|sem |0.8871|± |0.0162| | |
| |mm\|arc_easy\|0 | 0|sem |0.9377|± |0.0079| | |
| |mm\|commonsenseqa\|0| 0|sem |0.8031|± |0.0223| | |
| |mm\|gsm8k\|0 | 0|math_pass@1:1_samples|0.8993|± |0.0142| | |
| |mm\|math_500\|0 | 3|math_pass@1:1_samples|0.9750|± |0.0250| | |
| |mm\|truthfulqa\|0 | 0|sem |0.7107|± |0.0414| | |