Hugging Face's logo

Buckets:
Mercity
/
Desicion-Model-Custom

Files
xet
Mercity/Desicion-Model-Custom / run1 /logs /check_async.log
Pranav2748's picture
Pranav2748
download
raw
24.7 kB
(MainProcess pid=15023) INFO 10-31 00:24:19 [model.py:733] Resolved architecture: Qwen3_5ForConditionalGeneration
(MainProcess pid=15023) INFO 10-31 00:24:19 [model.py:2152] Using max model len 32768
Fetching 12 files: 0%| | 0/12 [00:00<?, ?it/s] Fetching 12 files: 100%|██████████| 12/12 [00:00<00:00, 995.92it/s]
(MainProcess pid=15023) INFO 10-31 00:24:21 [scheduler.py:307] Chunked prefill is enabled with max_num_batched_tokens=2048.
(MainProcess pid=15023) INFO 10-31 00:24:21 [config.py:688] Mamba cache mode is set to 'align' for Qwen3_5ForConditionalGeneration by default when prefix caching is enabled
Parse safetensors files: 0%| | 0/1 [00:00<?, ?it/s] Parse safetensors files: 100%|██████████| 1/1 [00:00<00:00, 4.96it/s] Parse safetensors files: 100%|██████████| 1/1 [00:00<00:00, 4.96it/s]
(MainProcess pid=15023) INFO 10-31 00:24:22 [kernel.py:467] Final IR op priority after setting platform defaults: IrOpPriorityConfig(rms_norm=['native'], fused_add_rms_norm=['native'], gelu_and_mul_sparse=['triton', 'native'])
(MainProcess pid=15023) WARNING 10-31 00:24:33 [system_utils.py:153] We must use the `spawn` multiprocessing start method. Overriding VLLM_WORKER_MULTIPROC_METHOD to 'spawn'. See https://docs.vllm.ai/en/latest/usage/troubleshooting.html#python-multiprocessing for more information. Reasons: CUDA is initialized
(EngineCore pid=15436) INFO 10-31 00:24:44 [core.py:133] Initializing a V1 LLM engine (v0.32.0) with config: model='Qwen/Qwen3.5-0.8B', speculative_config=None, tokenizer='Qwen/Qwen3.5-0.8B', skip_tokenizer_init=False, tokenizer_mode=auto, revision=main, tokenizer_revision=main, trust_remote_code=False, dtype=torch.bfloat16, max_seq_len=32768, download_dir=None, load_format=auto, tensor_parallel_size=1, pipeline_parallel_size=1, data_parallel_size=1, decode_context_parallel_size=1, dcp_comm_backend=ag_rs, disable_custom_all_reduce=False, quantization=None, quantization_config=None, enforce_eager=False, aux_output_config=AuxOutputConfig(enable_return_routed_experts=False, max_bytes=None), kv_cache_dtype=auto, device_config=cuda, structured_outputs_config=StructuredOutputsConfig(backend='auto', disable_any_whitespace=False, disable_additional_properties=False, reasoning_parser='', reasoning_parser_plugin='', enable_in_reasoning=False), observability_config=ObservabilityConfig(show_hidden_metrics_for_version=None, otlp_traces_endpoint=None, collect_detailed_traces=None, per_request_spec_decode_metrics='none', kv_cache_metrics=False, kv_cache_metrics_sample=0.01, custom_histogram_buckets=None, cudagraph_metrics=False, enable_layerwise_nvtx_tracing=False, enable_mfu_metrics=False, enable_mm_processor_stats=False, enable_logging_iteration_details=False, jit_monitor_mode='warn', jit_monitor_verbose=False), seed=17, served_model_name=Qwen/Qwen3.5-0.8B, enable_prefix_caching=True, enable_chunked_prefill=True, pooler_config=None, compilation_config={'mode': <CompilationMode.VLLM_COMPILE: 3>, 'debug_dump_path': None, 'cache_dir': '', 'compile_cache_save_format': 'binary', 'backend': 'inductor', 'custom_ops': ['none'], 'ir_enable_torch_wrap': True, 'splitting_ops': ['vllm::unified_attention_with_output', 'vllm::unified_mla_attention_with_output', 'vllm::mamba_mixer2', 'vllm::mamba_mixer', 'vllm::short_conv', 'vllm::qwen4_exp_ple_short_conv', 'vllm::qwen4_exp_qsa_with_output', 'vllm::qwen4_exp_qsa_prepare_with_output', 'vllm::linear_attention', 'vllm::qwen_gdn_attention_core', 'vllm::qwen_gdn_attention_core_fused_norm_packed', 'vllm::gdn_attention_core_xpu', 'vllm::olmo_hybrid_gdn_full_forward', 'vllm::sparse_attn_indexer', 'vllm::rocm_aiter_sparse_attn_indexer', 'vllm::deepseek_v4_attention', 'vllm::hpc_rope_norm_forward', 'vllm::unified_kv_cache_update', 'vllm::unified_mla_kv_cache_update'], 'compile_mm_encoder': False, 'cudagraph_mm_encoder': False, 'encoder_cudagraph_token_budgets': [], 'encoder_cudagraph_max_vision_items_per_batch': 0, 'encoder_cudagraph_max_frames_per_batch': None, 'cudagraph_decoder_replay': True, 'decoder_replay_cudagraph_capture_sizes': [], 'decoder_replay_trim_threshold': 768, 'compile_sizes': [], 'compile_ranges_endpoints': [2048], 'inductor_compile_config': {'enable_auto_functionalized_v2': False, 'combo_kernels': True, 'benchmark_combo_kernel': True}, 'inductor_passes': {}, 'cudagraph_mode': <CUDAGraphMode.FULL_AND_PIECEWISE: (2, 1)>, 'cudagraph_num_of_warmups': 1, 'cudagraph_capture_sizes': [1, 2, 4, 8, 16, 24, 32, 40, 48, 56, 64, 72, 80, 88, 96, 104, 112, 120, 128, 136, 144, 152, 160, 168, 176, 184, 192, 200, 208, 216, 224, 232, 240, 248, 256], 'cudagraph_copy_inputs': False, 'cudagraph_specialize_lora': True, 'use_inductor_graph_partition': False, 'pass_config': {'fuse_norm_quant': False, 'fuse_act_quant': False, 'fuse_attn_quant': False, 'fuse_allreduce_rms': False, 'enable_qk_norm_rope_fusion': False, 'fuse_rope_kvcache_cat_mla': False, 'fuse_act_padding': False, 'fuse_qk_norm_rope_kvcache': False}, 'max_cudagraph_capture_size': 256, 'dynamic_shapes_config': {'type': <DynamicShapesType.BACKED: 'backed'>, 'evaluate_guards': False, 'assume_32_bit_indexing': False}, 'local_cache_dir': None, 'fast_moe_cold_start': False, 'static_all_moe_layers': []}, kernel_config=KernelConfig(ir_op_priority=IrOpPriorityConfig(rms_norm=['native'], fused_add_rms_norm=['native'], gelu_and_mul_sparse=['triton', 'native']), enable_flashinfer_autotune=True, enable_rocm_segmented_attn_autotune=False, enable_cutedsl_warmup=True, enable_jit_warmup=True, moe_backend='auto', sparse_indexer_topk_backend='auto', linear_backend='auto', linear_backend_per_quant=None)
(EngineCore pid=15436) INFO 10-31 00:24:45 [parallel_state.py:1906] world_size=1 rank=0 local_rank=0 distributed_init_method=file:///tmp/vllm_dist_8119682ab1834e5487109c0df17babff backend=nccl
(EngineCore pid=15436) INFO 10-31 00:24:45 [parallel_state.py:2350] rank 0 in world size 1 is assigned as DP rank 0, PP rank 0, PCP rank 0, TP rank 0, ETP rank 0, EP rank N/A, EPLB rank N/A
(EngineCore pid=15436) INFO 10-31 00:24:45 [gpu_worker.py:518] Using V2 Model Runner
(EngineCore pid=15436) INFO 10-31 00:24:46 [model_runner.py:392] Loading model from scratch...
Fetching 12 files: 0%| | 0/12 [00:00<?, ?it/s] Fetching 12 files: 100%|██████████| 12/12 [00:00<00:00, 1027.18it/s]
(EngineCore pid=15436) INFO 10-31 00:24:47 [cuda.py:565] Using backend AttentionBackendEnum.FLASH_ATTN for vit attention
(EngineCore pid=15436) INFO 10-31 00:24:47 [mm_encoder_attention.py:372] Using AttentionBackendEnum.FLASH_ATTN for MMEncoderAttention.
(EngineCore pid=15436) INFO 10-31 00:24:47 [qwen_gdn_linear_attn.py:221] Using Triton/FLA GDN prefill kernel (requested=auto, head_k_dim=128).
(EngineCore pid=15436) INFO 10-31 00:24:48 [cuda.py:506] Using FLASH_ATTN attention backend out of potential backends: ['FLASH_ATTN', 'FLASHINFER', 'TRITON_ATTN', 'FLEX_ATTENTION'].
(EngineCore pid=15436) INFO 10-31 00:24:48 [flash_attn.py:1117] Using FlashAttention version 2
(EngineCore pid=15436) INFO 10-31 00:24:49 [base_loader.py:82] Initializing model took 2.141403 seconds
(EngineCore pid=15436) INFO 10-31 00:24:49 [weight_utils.py:1108] Filesystem type for checkpoints: XFS. Checkpoint size: 1.63 GiB. Available RAM: 196.80 GiB.
(EngineCore pid=15436) INFO 10-31 00:24:49 [weight_utils.py:1131] Auto-prefetch is disabled because the filesystem (XFS) is not a recognized network FS (NFS/Lustre). If you want to force prefetching, start vLLM with --safetensors-load-strategy=prefetch.
Loading safetensors checkpoint shards: 0% Completed | 0/1 [00:00<?, ?it/s]
(EngineCore pid=15436) WARNING 10-31 00:24:49 [utils.py:333] Attempted to load weight layers.0.linear_attn.norm.weight with dtype torch.float32 into parameter with dtype torch.bfloat16
(EngineCore pid=15436) WARNING 10-31 00:24:49 [utils.py:333] Attempted to load weight layers.1.linear_attn.norm.weight with dtype torch.float32 into parameter with dtype torch.bfloat16
(EngineCore pid=15436) WARNING 10-31 00:24:49 [utils.py:333] Attempted to load weight layers.10.linear_attn.norm.weight with dtype torch.float32 into parameter with dtype torch.bfloat16
(EngineCore pid=15436) WARNING 10-31 00:24:49 [utils.py:333] Attempted to load weight layers.12.linear_attn.norm.weight with dtype torch.float32 into parameter with dtype torch.bfloat16
(EngineCore pid=15436) WARNING 10-31 00:24:49 [utils.py:333] Attempted to load weight layers.13.linear_attn.norm.weight with dtype torch.float32 into parameter with dtype torch.bfloat16
(EngineCore pid=15436) WARNING 10-31 00:24:49 [utils.py:333] Attempted to load weight layers.14.linear_attn.norm.weight with dtype torch.float32 into parameter with dtype torch.bfloat16
(EngineCore pid=15436) WARNING 10-31 00:24:49 [utils.py:333] Attempted to load weight layers.16.linear_attn.norm.weight with dtype torch.float32 into parameter with dtype torch.bfloat16
(EngineCore pid=15436) WARNING 10-31 00:24:49 [utils.py:333] Attempted to load weight layers.17.linear_attn.norm.weight with dtype torch.float32 into parameter with dtype torch.bfloat16
(EngineCore pid=15436) WARNING 10-31 00:24:49 [utils.py:333] Attempted to load weight layers.18.linear_attn.norm.weight with dtype torch.float32 into parameter with dtype torch.bfloat16
(EngineCore pid=15436) WARNING 10-31 00:24:49 [utils.py:333] Attempted to load weight layers.2.linear_attn.norm.weight with dtype torch.float32 into parameter with dtype torch.bfloat16
(EngineCore pid=15436) WARNING 10-31 00:24:49 [utils.py:333] Attempted to load weight layers.20.linear_attn.norm.weight with dtype torch.float32 into parameter with dtype torch.bfloat16
(EngineCore pid=15436) WARNING 10-31 00:24:49 [utils.py:333] Attempted to load weight layers.21.linear_attn.norm.weight with dtype torch.float32 into parameter with dtype torch.bfloat16
(EngineCore pid=15436) WARNING 10-31 00:24:49 [utils.py:333] Attempted to load weight layers.22.linear_attn.norm.weight with dtype torch.float32 into parameter with dtype torch.bfloat16
(EngineCore pid=15436) WARNING 10-31 00:24:49 [utils.py:333] Attempted to load weight layers.4.linear_attn.norm.weight with dtype torch.float32 into parameter with dtype torch.bfloat16
(EngineCore pid=15436) WARNING 10-31 00:24:49 [utils.py:333] Attempted to load weight layers.5.linear_attn.norm.weight with dtype torch.float32 into parameter with dtype torch.bfloat16
(EngineCore pid=15436) WARNING 10-31 00:24:49 [utils.py:333] Attempted to load weight layers.6.linear_attn.norm.weight with dtype torch.float32 into parameter with dtype torch.bfloat16
(EngineCore pid=15436) WARNING 10-31 00:24:49 [utils.py:333] Attempted to load weight layers.8.linear_attn.norm.weight with dtype torch.float32 into parameter with dtype torch.bfloat16
(EngineCore pid=15436) WARNING 10-31 00:24:49 [utils.py:333] Attempted to load weight layers.9.linear_attn.norm.weight with dtype torch.float32 into parameter with dtype torch.bfloat16
Loading safetensors checkpoint shards: 100% Completed | 1/1 [00:00<00:00, 4.37it/s]
Loading safetensors checkpoint shards: 100% Completed | 1/1 [00:00<00:00, 4.37it/s]
(EngineCore pid=15436) INFO 10-31 00:24:49 [default_loader.py:501] Loading weights took 0.23 seconds
(EngineCore pid=15436) INFO 10-31 00:24:50 [model_runner.py:424] Model loading took 1.63 GiB memory and 4.186301 seconds
(EngineCore pid=15436) INFO 10-31 00:24:52 [topk_topp_sampler.py:124] Using FlashInfer for top-p & top-k sampling.
(EngineCore pid=15436) INFO 10-31 00:24:52 [torch_utils.py:295] Reducing Torch threads from 128 to 1 for serving. Set OMP_NUM_THREADS in the external environment to override.
(EngineCore pid=15436) INFO 10-31 00:24:52 [interface.py:1096] Setting attention block size to 544 tokens to ensure that attention page size is >= mamba page size.
(EngineCore pid=15436) INFO 10-31 00:24:52 [interface.py:1120] Padding mamba page size by 2.64% to ensure that mamba page size and attention page size are exactly equal.
(EngineCore pid=15436) INFO 10-31 00:24:52 [utils.py:347] Using LBNHC KV cache layout.
[transformers] Qwen3VL video processing does not apply the per-frame pixel cap the reference implementation (qwen-vl-utils) applies, so some videos cost far more tokens than they would there. In v5.22 the capped behavior will become the default and `cap_pixels_per_frame` will be removed. Pass `cap_pixels_per_frame=True` to adopt the reference behavior now, or `False` to keep the current behavior and silence this warning.
(MainProcess pid=15023) INFO 10-31 00:24:53 [base.py:280] Multi-modal warmup completed in 20.264s
(EngineCore pid=15436) INFO 10-31 00:25:00 [encoder_runner.py:131] Encoder cache will be initialized with a budget of 16384 tokens, and profiled with 1 image items of the maximum feature size.
(EngineCore pid=15436) INFO 10-31 00:25:02 [caching.py:342] reconstructed serializable fn from standalone compile artifacts. num_artifacts=11 num_submods=25
(EngineCore pid=15436) INFO 10-31 00:25:02 [decorators.py:319] [backbone] Directly load AOT compilation from path /root/.cache/vllm/torch_compile_cache/torch_aot_compile/38b27199840dd5788649ecfb9a7fd4945e985542636a6f138f09557d6ee82c7d/rank_0_0/model
(EngineCore pid=15436) INFO 10-31 00:25:02 [monitor.py:62] [backbone] torch.compile took 0.80 s in total
(EngineCore pid=15436) INFO 10-31 00:25:03 [monitor.py:91] [backbone] Initial profiling/warmup run took 0.26 s
Capturing CUDA graphs (PIECEWISE): 0%| | 0/35 [00:00<?, ?it/s] Capturing CUDA graphs (PIECEWISE): 3%|▎ | 1/35 [00:01<00:54, 1.61s/it] Capturing CUDA graphs (PIECEWISE): 6%|▌ | 2/35 [00:01<00:25, 1.27it/s] Capturing CUDA graphs (PIECEWISE): 9%|▊ | 3/35 [00:02<00:16, 1.92it/s] Capturing CUDA graphs (PIECEWISE): 11%|█▏ | 4/35 [00:02<00:12, 2.55it/s] Capturing CUDA graphs (PIECEWISE): 14%|█▍ | 5/35 [00:02<00:10, 2.98it/s] Capturing CUDA graphs (PIECEWISE): 17%|█▋ | 6/35 [00:02<00:08, 3.59it/s] Capturing CUDA graphs (PIECEWISE): 20%|██ | 7/35 [00:02<00:06, 4.14it/s] Capturing CUDA graphs (PIECEWISE): 23%|██▎ | 8/35 [00:02<00:05, 4.70it/s] Capturing CUDA graphs (PIECEWISE): 26%|██▌ | 9/35 [00:03<00:04, 5.29it/s] Capturing CUDA graphs (PIECEWISE): 29%|██▊ | 10/35 [00:03<00:04, 5.74it/s] Capturing CUDA graphs (PIECEWISE): 31%|███▏ | 11/35 [00:03<00:03, 6.26it/s] Capturing CUDA graphs (PIECEWISE): 34%|███▍ | 12/35 [00:03<00:03, 6.61it/s] Capturing CUDA graphs (PIECEWISE): 37%|███▋ | 13/35 [00:03<00:03, 7.14it/s] Capturing CUDA graphs (PIECEWISE): 40%|████ | 14/35 [00:03<00:02, 7.75it/s] Capturing CUDA graphs (PIECEWISE): 43%|████▎ | 15/35 [00:03<00:02, 8.30it/s] Capturing CUDA graphs (PIECEWISE): 49%|████▊ | 17/35 [00:03<00:01, 10.08it/s] Capturing CUDA graphs (PIECEWISE): 57%|█████▋ | 20/35 [00:04<00:01, 14.78it/s] Capturing CUDA graphs (PIECEWISE): 66%|██████▌ | 23/35 [00:04<00:00, 18.65it/s] Capturing CUDA graphs (PIECEWISE): 74%|███████▍ | 26/35 [00:04<00:00, 21.42it/s] Capturing CUDA graphs (PIECEWISE): 83%|████████▎ | 29/35 [00:04<00:00, 19.26it/s] Capturing CUDA graphs (PIECEWISE): 91%|█████████▏| 32/35 [00:04<00:00, 21.78it/s] Capturing CUDA graphs (PIECEWISE): 100%|██████████| 35/35 [00:04<00:00, 22.78it/s] Capturing CUDA graphs (PIECEWISE): 100%|██████████| 35/35 [00:04<00:00, 7.51it/s]
Capturing CUDA graphs (FULL): 0%| | 0/2 [00:00<?, ?it/s] Capturing CUDA graphs (FULL): 100%|██████████| 2/2 [00:00<00:00, 24.11it/s]
(EngineCore pid=15436) INFO 10-31 00:25:10 [model_runner.py:1127] Graph capturing finished in 6 secs, took 0.21 GiB
(EngineCore pid=15436) INFO 10-31 00:25:10 [gpu_worker.py:759] Available KV cache memory: 67.94 GiB
(EngineCore pid=15436) INFO 10-31 00:25:10 [gpu_worker.py:774] CUDA graph memory profiling is enabled (default since v0.21.0). The current --gpu-memory-utilization=0.9000 is equivalent to --gpu-memory-utilization=0.8972 without CUDA graph memory profiling. To maintain the same effective KV cache size as before, increase --gpu-memory-utilization to 0.9028. To disable, set VLLM_MEMORY_PROFILER_ESTIMATE_CUDAGRAPHS=0.
(EngineCore pid=15436) INFO 10-31 00:25:10 [kv_cache_utils.py:2327] GPU KV cache size: 5,336,782 tokens, Maximum concurrency for 32,768 tokens per request: 162.87x
(EngineCore pid=15436) INFO 10-31 00:25:10 [kv_cache_utils.py:853] kv cache group sizes (544, 544, 544, 544)
(EngineCore pid=15436) INFO 10-31 00:25:10 [kv_cache_utils.py:854] kv lcm block sizes 544
(EngineCore pid=15436) INFO 10-31 00:25:10 [kernel_warmup.py:176] JIT kernel warmup starting.
(EngineCore pid=15436) INFO 10-31 00:25:10 [kernel_warmup.py:189] JIT kernel warmup finished in 0.00s.
(EngineCore pid=15436) INFO 10-31 00:25:10 [qwen_triton_warmup.py:296] Warming up Qwen GDN Triton kernels for model_type=qwen3_5_text.
(EngineCore pid=15436) INFO 10-31 00:25:10 [qwen_vl_triton_warmup.py:57] Warmed position embedding and vision rotary kernels on grids=[(1, 16, 16), (1, 16, 2), (1, 2, 16), (1, 2, 2)].
(EngineCore pid=15436) INFO 10-31 00:25:11 [mamba_triton_warmup.py:42] Warmed Mamba batch_memcpy_kernel.
Capturing CUDA graphs (PIECEWISE): 0%| | 0/35 [00:00<?, ?it/s] Capturing CUDA graphs (PIECEWISE): 3%|▎ | 1/35 [00:00<00:11, 3.06it/s] Capturing CUDA graphs (PIECEWISE): 6%|▌ | 2/35 [00:00<00:09, 3.52it/s] Capturing CUDA graphs (PIECEWISE): 9%|▊ | 3/35 [00:00<00:07, 4.13it/s] Capturing CUDA graphs (PIECEWISE): 11%|█▏ | 4/35 [00:01<00:07, 4.00it/s] Capturing CUDA graphs (PIECEWISE): 14%|█▍ | 5/35 [00:01<00:06, 4.50it/s] Capturing CUDA graphs (PIECEWISE): 17%|█▋ | 6/35 [00:01<00:06, 4.40it/s] Capturing CUDA graphs (PIECEWISE): 20%|██ | 7/35 [00:01<00:05, 4.90it/s] Capturing CUDA graphs (PIECEWISE): 23%|██▎ | 8/35 [00:01<00:05, 4.94it/s] Capturing CUDA graphs (PIECEWISE): 26%|██▌ | 9/35 [00:01<00:04, 5.33it/s] Capturing CUDA graphs (PIECEWISE): 29%|██▊ | 10/35 [00:02<00:04, 5.90it/s] Capturing CUDA graphs (PIECEWISE): 31%|███▏ | 11/35 [00:02<00:04, 5.54it/s] Capturing CUDA graphs (PIECEWISE): 34%|███▍ | 12/35 [00:02<00:03, 6.20it/s] Capturing CUDA graphs (PIECEWISE): 37%|███▋ | 13/35 [00:02<00:03, 6.91it/s] Capturing CUDA graphs (PIECEWISE): 40%|████ | 14/35 [00:02<00:03, 6.52it/s] Capturing CUDA graphs (PIECEWISE): 43%|████▎ | 15/35 [00:02<00:02, 7.27it/s] Capturing CUDA graphs (PIECEWISE): 49%|████▊ | 17/35 [00:02<00:01, 9.96it/s] Capturing CUDA graphs (PIECEWISE): 60%|██████ | 21/35 [00:03<00:00, 16.47it/s] Capturing CUDA graphs (PIECEWISE): 66%|██████▌ | 23/35 [00:03<00:00, 16.80it/s] Capturing CUDA graphs (PIECEWISE): 77%|███████▋ | 27/35 [00:03<00:00, 21.53it/s] Capturing CUDA graphs (PIECEWISE): 89%|████████▊ | 31/35 [00:03<00:00, 25.02it/s] Capturing CUDA graphs (PIECEWISE): 100%|██████████| 35/35 [00:03<00:00, 27.72it/s] Capturing CUDA graphs (PIECEWISE): 100%|██████████| 35/35 [00:03<00:00, 10.01it/s]
Capturing CUDA graphs (FULL): 0%| | 0/19 [00:00<?, ?it/s] Capturing CUDA graphs (FULL): 16%|█▌ | 3/19 [00:00<00:00, 28.64it/s] Capturing CUDA graphs (FULL): 37%|███▋ | 7/19 [00:00<00:00, 34.20it/s] Capturing CUDA graphs (FULL): 58%|█████▊ | 11/19 [00:00<00:00, 35.98it/s] Capturing CUDA graphs (FULL): 79%|███████▉ | 15/19 [00:00<00:00, 30.75it/s] Capturing CUDA graphs (FULL): 100%|██████████| 19/19 [00:00<00:00, 33.55it/s] Capturing CUDA graphs (FULL): 100%|██████████| 19/19 [00:00<00:00, 33.16it/s]
(EngineCore pid=15436) INFO 10-31 00:25:16 [model_runner.py:1127] Graph capturing finished in 5 secs, took 0.12 GiB
(EngineCore pid=15436) INFO 10-31 00:25:16 [gpu_worker.py:971] CUDA graph pool memory: 0.12 GiB (actual), 0.22 GiB (estimated), difference: 0.11 GiB (88.5%).
(EngineCore pid=15436) INFO 10-31 00:25:16 [gpu_worker.py:1034] Free memory on device (78.84/79.25 GiB) on startup. Desired GPU memory utilization is (0.9, 71.33 GiB). Actual usage is 1.94 GiB for consumed memory (weights + non-torch), 1.45 GiB for peak activation, and 0.12 GiB for CUDAGraph memory. Replace gpu_memory_utilization config with `--kv-cache-memory=72659490714` (67.67 GiB) to fit into requested memory, or `--kv-cache-memory=80725556736` (75.18 GiB) to fully utilize gpu memory. Current kv cache memory in use is 67.94 GiB.
(EngineCore pid=15436) INFO 10-31 00:25:16 [jit_monitor.py:84] Kernel JIT monitor activated; monitored JIT compilations during inference will use mode=warn.
(EngineCore pid=15436) INFO 10-31 00:25:17 [core.py:395] init engine (profile, create kv cache, warmup model) took 24.60 s (compilation: 0.80 s)
Parse safetensors files: 0%| | 0/1 [00:00<?, ?it/s] Parse safetensors files: 100%|██████████| 1/1 [00:00<00:00, 3.48it/s] Parse safetensors files: 100%|██████████| 1/1 [00:00<00:00, 3.48it/s]
(EngineCore pid=15436) INFO 10-31 00:25:17 [kernel.py:467] Final IR op priority after setting platform defaults: IrOpPriorityConfig(rms_norm=['native'], fused_add_rms_norm=['native'], gelu_and_mul_sparse=['triton', 'native'])
(MainProcess pid=15023) WARNING 10-31 00:25:17 [input_processor.py:408] Passing raw prompts to InputProcessor is deprecated and will be removed in the future. You should instead pass the outputs of Renderer.render_cmpl() or Renderer.render_chat().
(EngineCore pid=15436) WARNING 10-31 00:25:18 [jit_monitor.py:140] Triton kernel JIT compilation during inference: _topp_sb_stats_kernel. This causes a latency spike; consider extending warmup to cover this shape/config.
(EngineCore pid=15436) WARNING 10-31 00:25:18 [jit_monitor.py:140] Triton kernel JIT compilation during inference: _topp_sb_step_kernel. This causes a latency spike; consider extending warmup to cover this shape/config.
(EngineCore pid=15436) WARNING 10-31 00:25:18 [jit_monitor.py:140] Triton kernel JIT compilation during inference: _topp_sb_mask_kernel. This causes a latency spike; consider extending warmup to cover this shape/config.
progress {"time": "00:26:17", "elapsed_min": 1.0, "submitted": 4, "finished": 0, "in_flight": 4, "gen_tokens": 0, "tok_per_s": 0, "chain_tokens_median": 0.0, "chain_tokens_p90": 0.0, "chain_tokens_max": 0, "parse_rate": null, "looped": 0, "minutes_to_deadline": Infinity}
progress {"time": "00:27:17", "elapsed_min": 2.0, "submitted": 4, "finished": 0, "in_flight": 4, "gen_tokens": 0, "tok_per_s": 0, "chain_tokens_median": 0.0, "chain_tokens_p90": 0.0, "chain_tokens_max": 0, "parse_rate": null, "looped": 0, "minutes_to_deadline": Infinity}
(MainProcess pid=15023) INFO 10-31 00:27:58 [core_client.py:770] [shutdown] MPClient: start timeout=default
(MainProcess pid=15023) INFO 10-31 00:27:58 [core_client.py:772] [shutdown] MPClient: stopping engine manager
(MainProcess pid=15023) INFO 10-31 00:27:58 [utils.py:628] [shutdown] Process manager: send sigterm to process EngineCore
(EngineCore pid=15436) INFO 10-31 00:27:58 [core.py:1454] [shutdown] EngineCore: trigger received signal=SIGTERM
(EngineCore pid=15436) INFO 10-31 00:27:58 [core.py:1635] [shutdown] EngineCore: start mode=abort timeout=0s
(EngineCore pid=15436) INFO 10-31 00:27:58 [core.py:1666] [shutdown] EngineCore: request processing complete; starting resource teardown
(EngineCore pid=15436) INFO 10-31 00:27:58 [core.py:1499] [shutdown] EngineCore: exiting busy loop
(MainProcess pid=15023) INFO 10-31 00:28:02 [core_client.py:774] [shutdown] MPClient: engine manager stopped
(MainProcess pid=15023) INFO 10-31 00:28:02 [core_client.py:775] [shutdown] MPClient: cleaning up background resources
(MainProcess pid=15023) INFO 10-31 00:28:02 [core_client.py:777] [shutdown] MPClient: complete
{
"items": 4,
"samples": 16,
"tokens_total": 419664,
"tokens_median": 32211.0,
"tokens_p90": 32334.0,
"tokens_max": 32334,
"parse_rate": 0.1875,
"looped": 13,
"items_with_correct_chain": 1,
"majority_accuracy": 0.25,
"seconds": 223.25800108909607,
"gen_tok_per_s": 1879.7265851740908,
"engine": {
"submitted": 4,
"finished": 4,
"gen_tokens": 419664,
"t0": 1793406317.22621,
"seconds": 161.5282564163208,
"tok_per_s": 2598.0841328365705
},
"N_fit": 177
}

Xet Storage Details

Size:
24.7 kB
·
Xet hash:
251dc88415ac8cee3e1e7303a6d45ef57a58ab1ca43abcb2d0e6b277ae39daf4

Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.