[2026-10-17 15:40:02] Registered linear attn model: config=GigaChat35Config, backend=GDNAttnBackend, archs=['GigaChat35ForCausalLM', 'GigaChat35ForCausalLMNextN']
[2026-10-17 15:40:03] Failed to get GPU memory capacity from nvidia-smi. Falling back to torch.cuda.mem_get_info(). Reported total GPU memory per device (MiB): [124610], using min: 124610 MiB.
[2026-10-17 15:40:03] FlashInfer allreduce fusion is forcibly disabled via --enforce-disable-flashinfer-allreduce-fusion.
[2026-10-17 15:40:03] Sampling backend is set to pytorch for deterministic inference.
[2026-10-17 15:40:03] Attention backend not specified. Falling back to 'flashinfer' for deterministic inference. You can explicitly set --attention-backend to one of ['ascend', 'fa3', 'fa4', 'flashinfer', 'intel_xpu', 'triton'].
[2026-10-17 15:40:03] Currently radix cache is not compatible with flashinfer attention backend for deterministic inference. It will be supported in the future.
[2026-10-17 15:40:04] server_args={'model_path': 'Qwen/Qwen3-30B-A3B-Instruct-2507', 'tokenizer_path': 'Qwen/Qwen3-30B-A3B-Instruct-2507', 'tokenizer_mode': 'auto', 'tokenizer_backend': 'huggingface', 'tokenizer_worker_num': 1, 'detokenizer_worker_num': 1, 'skip_tokenizer_init': False, 'load_format': 'auto', 'model_loader_extra_config': '{}', 'trust_remote_code': False, 'context_length': None, 'is_embedding': False, 'enable_multimodal': None, 'revision': None, 'model_impl': 'auto', 'model_config_parser': 'auto', 'json_model_override_args': '{}', 'dtype': 'auto', 'quantization': None, 'quantization_param_path': None, 'kv_cache_dtype': 'auto', 'enable_fp32_lm_head': False, 'modelopt_quant': None, 'modelopt_checkpoint_restore_path': None, 'modelopt_checkpoint_save_path': None, 'modelopt_export_path': None, 'quantize_and_serve': False, 'rl_quant_profile': None, 'enable_tf32_matmul': False, 'mem_fraction_static': 0.7, 'max_running_requests': None, 'max_queued_requests': None, 'max_total_tokens': None, 'chunked_prefill_size': 8192, 'prefill_decode_interval': 0, 'enable_dynamic_chunking': False, 'max_prefill_tokens': 16384, 'prefill_max_requests': None, 'schedule_policy': 'fcfs', 'enable_priority_scheduling': False, 'disable_priority_preemption': False, 'default_priority_value': None, 'abort_on_priority_when_disabled': False, 'schedule_low_priority_values_first': False, 'priority_scheduling_preemption_threshold': 10, 'retraction_policy': 'length', 'schedule_conservativeness': 1.0, 'page_size': 1, 'c128_page_size': 16, 'swa_full_tokens_ratio': 0.8, 'disable_hybrid_swa_memory': False, 'radix_eviction_policy': 'lru', 'radix_eviction_policy_config': None, 'prefill_only_disable_kv_cache': False, 'disable_radix_cache': True, 'enable_page_major_kv_layout': False, 'enable_unified_memory': False, 'disable_chunked_prefix_cache': False, 'disable_overlap_schedule': False, 'num_continuous_decode_steps': 1, 'scheduler_recv_interval': 1, 'enable_mixed_chunk': False, 'nccl_port': None, 'dist_timeout': None, 'dist_init_addr': None, 'gated_launch_port': None, 'nnodes': 1, 'node_rank': 0, 'tp_size': 1, 'dcp_size': 1, 'pp_size': 1, 'pp_max_micro_batch_size': None, 'pp_async_batch_depth': 0, 'dp_size': 1, 'attn_dp_size': 1, 'load_balance_method': 'round_robin', 'attn_cp_size': 1, 'moe_dp_size': 1, 'dwdp_size': 1, 'dcp_comm_backend': 'ag_rs', 'dcp_replicate_q_proj': None, 'enable_prefill_cp': False, 'cp_strategy': None, 'enable_dsa_cache_layer_split': False, 'enable_cp_decode_attn_tp': False, 'enable_kv_cache_sharding': False, 'enable_dp_attention': False, 'enable_dp_attention_local_control_broadcast': False, 'enable_dp_lm_head': False, 'enable_tp_lm_head_all_to_all': False, 'enable_attn_tp_input_scattered': False, 'enable_shared_experts_attn_tp': False, 'shared_experts_tp_size': None, 'enable_dense_mlp_attn_tp': False, 'enable_layernorm_sp': False, 'disable_attn_tp_gather': False, 'enable_p2p_check': False, 'device': 'cuda', 'base_gpu_id': 0, 'gpu_id_step': 1, 'random_seed': 529614751, 'mlx_enable_sampling': False, 'watchdog_timeout': 300, 'soft_watchdog_timeout': None, 'sleep_on_idle': False, 'use_ray': False, 'custom_sigquit_handler': None, 'numa_node': None, 'gc_threshold': None, 'host': '127.0.0.1', 'port': 30000, 'fastapi_root_path': '', 'smg_grpc_mode': False, 'grpc_mode': False, 'grpc_port': None, 'grpc_worker_threads': 4, 'sidecar': None, 'sidecar_args': None, 'skip_server_warmup': False, 'warmups': None, 'enable_http2': False, 'http2_max_concurrent_streams': 200, 'http2_initial_connection_window_size': 1048576, 'ssl_keyfile': None, 'ssl_certfile': None, 'ssl_ca_certs': None, 'ssl_keyfile_password': None, 'enable_ssl_refresh': False, 'api_key': None, 'admin_api_key': None, 'served_model_name': 'Qwen/Qwen3-30B-A3B-Instruct-2507', 'weight_version': 'default', 'chat_template': None, 'trust_request_chat_template': False, 'hf_chat_template_name': None, 'completion_template': None, 'file_storage_path': None, 'enable_cache_report': True, 'return_input_ids': False, 'return_output_ids': False, 'reasoning_parser': None, 'default_chat_template_kwargs': None, 'strip_thinking_cache': False, 'enable_strict_thinking': False, 'tool_call_parser': None, 'tool_server': None, 'sampling_defaults': 'model', 'asr_max_buffer_seconds': 60, 'asr_max_concurrent_sessions': 32, 'preferred_sampling_params': None, 'allow_auto_truncate': False, 'stream_interval': 1, 'batch_notify_size': 16, 'stream_response_default_include_usage': False, 'incremental_streaming_output': False, 'enable_streaming_session': False, 'enable_session_radix_cache': False, 'log_level': 'info', 'log_level_http': None, 'log_requests': False, 'log_requests_level': 2, 'log_requests_format': 'text', 'log_requests_target': None, 'uvicorn_access_log_exclude_prefixes': [], 'crash_dump_folder': None, 'show_time_cost': False, 'enable_metrics': True, 'smg_http_sidecar_port': None, 'enable_mfu_metrics': False, 'enable_metrics_for_all_schedulers': False, 'load_snapshot_publish_interval': 15, 'tokenizer_metrics_custom_labels_header': 'x-custom-labels', 'tokenizer_metrics_allowed_custom_labels': None, 'extra_metric_labels': None, 'bucket_time_to_first_token': None, 'bucket_inter_token_latency': None, 'bucket_e2e_request_latency': None, 'prompt_tokens_buckets': None, 'generation_tokens_buckets': None, 'gc_warning_threshold_secs': 0.0, 'decode_log_interval': 40, 'enable_request_time_stats_logging': False, 'kv_events_config': None, 'load_publish_endpoint': None, 'enable_forward_pass_metrics': False, 'forward_pass_metrics_worker_id': '', 'forward_pass_metrics_ipc_name': None, 'enable_trace': False, 'trace_modules': 'request', 'otlp_traces_endpoint': 'localhost:4317', 'export_metrics_to_file': False, 'export_metrics_to_file_dir': None, 'stat_loggers': None, 'constrained_json_whitespace_pattern': None, 'constrained_json_disable_any_whitespace': False, 'constrained_json_max_whitespace_cnt': None, 'attention_backend': 'flashinfer', 'decode_attention_backend': None, 'enable_lean_attention': None, 'prefill_attention_backend': None, 'sampling_backend': 'pytorch', 'grammar_backend': 'xgrammar', 'radix_cache_backend': None, 'mm_attention_backend': None, 'fp8_gemm_runner_backend': 'auto', 'fp4_gemm_runner_backend': 'auto', 'bf16_gemm_backend': 'auto', 'dsa_prefill_backend': None, 'dsv4_prefill_backend': 'auto', 'dsa_decode_backend': None, 'dsa_paged_mqa_logits_backend': 'auto', 'dsa_topk_backend': 'sgl-kernel', 'disable_flashinfer_autotune': False, 'flashinfer_autotune_skip_ops': None, 'mamba_backend': 'triton', 'cuda_graph_config': {'decode': {'backend': 'full', 'max_bs': 256, 'bs': [1, 2, 4, 8, 12, 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], 'tc_compiler': 'eager', 'max_context_size': None, 'full_prefill_max_req': None, 'full_prefill_prefix_chunk_tokens': None, 'max_seq_len': None}, 'prefill': {'backend': 'breakable', 'max_bs': 8192, 'bs': [4, 8, 12, 16, 20, 24, 28, 32, 48, 64, 80, 96, 112, 128, 144, 160, 176, 192, 208, 224, 240, 256, 288, 320, 352, 384, 416, 448, 480, 512, 576, 640, 704, 768, 832, 896, 960, 1024, 1280, 1536, 1792, 2048, 2304, 2560, 2816, 3072, 3328, 3584, 3840, 4096, 4608, 5120, 5632, 6144, 6656, 7168, 7680, 8192], 'tc_compiler': 'eager', 'max_context_size': None, 'full_prefill_max_req': None, 'full_prefill_prefix_chunk_tokens': None, 'max_seq_len': None}}, 'cuda_graph_backend_decode': None, 'cuda_graph_backend_prefill': None, 'cuda_graph_max_bs_decode': None, 'cuda_graph_max_bs_prefill': None, 'cuda_graph_bs_decode': None, 'cuda_graph_bs_prefill': None, 'cuda_graph_prefill_max_context': None, 'disable_prefill_cuda_graph': False, 'disable_decode_cuda_graph': False, 'disable_cuda_graph': False, 'disable_cuda_graph_padding': False, 'enable_profile_cuda_graph': False, 'enable_cudagraph_gc': False, 'debug_cuda_graph': False, 'enable_layerwise_nvtx_marker': False, 'enable_nccl_nvls': False, 'enable_symm_mem': False, 'triton_attention_reduce_in_fp32': False, 'triton_attention_num_kv_splits': 8, 'triton_attention_split_tile_size': None, 'flashinfer_mla_disable_ragged': False, 'enable_fused_qk_norm_rope': False, 'enable_precise_embedding_interpolation': False, 'enable_fused_moe_sum_all_reduce': False, 'enable_deepseek_v4_fp4_indexer': False, 'disable_custom_all_reduce': False, 'enable_mscclpp': False, 'enable_torch_symm_mem': False, 'enable_scattered_sconv': False, 'pre_warm_nccl': False, 'boundary_reduction': 'rs+rsv', 'enable_quant_communications': False, 'enable_flashinfer_allreduce_fusion': False, 'enforce_disable_flashinfer_allreduce_fusion': True, 'flashinfer_allreduce_fusion_backend': None, 'enable_aiter_allreduce_fusion': False, 'enable_torch_compile': False, 'enable_torch_compile_debug_mode': False, 'torch_compile_max_bs': 32, 'speculative_algorithm': None, 'uno_lora_path': None, 'speculative_draft_model_path': None, 'speculative_draft_model_revision': None, 'speculative_draft_load_format': None, 'speculative_num_steps': None, 'speculative_eagle_topk': None, 'speculative_num_draft_tokens': None, 'speculative_dflash_block_size': None, 'speculative_dspark_block_size': None, 'speculative_dspark_sps_table_path': None, 'speculative_dspark_confidence_sts_path': None, 'speculative_dspark_align_verify_tokens_to_graph_tier': False, 'speculative_accept_threshold_single': 1.0, 'speculative_accept_threshold_acc': 1.0, 'speculative_use_rejection_sampling': False, 'speculative_token_map': None, 'speculative_attention_mode': 'prefill', 'speculative_draft_attention_backend': None, 'speculative_dsa_topk_backend': 'sgl-kernel', 'speculative_draft_kv_cache_dtype': None, 'speculative_draft_window_size': None, 'speculative_moe_runner_backend': 'auto', 'speculative_moe_a2a_backend': None, 'speculative_enable_w4a4_mxfp4_megamoe': None, 'speculative_draft_model_quantization': None, '_speculative_draft_quantization_explicitly_set': False, 'speculative_skip_dp_mlp_sync': False, 'enable_multi_layer_eagle': False, 'speculative_adaptive': False, 'speculative_adaptive_config': None, 'decoupled_spec_bind_endpoint': None, 'decoupled_spec_connect_endpoints': None, 'decoupled_spec_rank': None, 'decoupled_spec_role': 'null', 'spec_trace_dir': None, 'speculative_ngram_min_bfs_breadth': 1, 'speculative_ngram_max_bfs_breadth': 10, 'speculative_ngram_match_type': 'BFS', 'speculative_ngram_max_trie_depth': 18, 'speculative_ngram_capacity': 10000000, 'speculative_ngram_external_corpus_path': None, 'speculative_ngram_external_sam_budget': 0, 'speculative_ngram_external_corpus_max_tokens': 10000000, 'ep_size': 1, 'moe_a2a_backend': 'none', 'enable_w4a4_mxfp4_megamoe': False, 'deepep_v2_mode': 'direct', 'moe_runner_backend': 'auto', 'flashinfer_mxfp4_moe_precision': 'default', 'deepep_mode': 'auto', 'fuseep_mode': 2, 'deepep_dispatcher_output_dtype': 'auto', 'ep_num_redundant_experts': 0, 'ep_dispatch_algorithm': None, 'init_expert_location': 'trivial', 'enable_eplb': False, 'eplb_algorithm': 'auto', 'eplb_rebalance_num_iterations': 1000, 'eplb_rebalance_layers_per_chunk': None, 'eplb_min_rebalancing_utilization_threshold': 1.0, 'expert_distribution_recorder_mode': None, 'expert_distribution_recorder_buffer_size': 1000, 'expert_balancedness_report_mode': 'off', 'deepep_config': None, 'moe_dense_tp_size': None, 'elastic_ep_backend': None, 'enable_elastic_expert_backup': False, 'mooncake_ib_device': None, 'enable_waterfill': False, 'ep_join_mode': None, 'ep_join_rank_offset': 0, 'elastic_ep_initial_size': None, 'max_ep_size': None, 'elastic_ep_scale_timeout': 600, 'disable_flashinfer_cutlass_moe_fp4_allgather': False, 'disable_shared_experts_fusion': False, 'enforce_shared_experts_fusion': False, 'max_mamba_cache_size': None, 'mamba_ssm_dtype': None, 'mamba_max_states_per_path': -1, 'enable_mamba_cache_stochastic_rounding': False, 'mamba_cache_philox_rounds': 0, 'mamba_full_memory_ratio': 0.9, 'mamba_radix_cache_strategy': 'auto', 'uses_mamba_radix_cache': False, 'mamba_track_interval': 256, 'enable_int8_mamba_checkpoint': False, 'int8_mamba_ckpt_size': None, 'linear_attn_backend': 'triton', 'linear_attn_decode_backend': None, 'linear_attn_prefill_backend': None, 'linear_attn_verify_backend': None, 'enable_linear_replayssm': False, 'linear_replayssm_cache_len': 16, 'enable_linear_replayssm_spec': False, 'enable_hierarchical_cache': False, 'hicache_host_memory_mode': 'cache', 'hicache_ratio': 2.0, 'hicache_size': 0, 'hicache_write_policy': 'write_through', 'hicache_io_backend': 'kernel', 'hicache_mem_layout': 'page_first', 'hicache_storage_backend': None, 'hicache_storage_prefetch_policy': 'timeout', 'hicache_storage_backend_extra_config': None, 'hicache_storage_prefetch_retry_poll_interval': 8, 'hicache_storage_prefetch_retry_max_attempts': 8, 'enable_unified_cache_external_linker': False, 'unified_cache_external_linker_backend': 'mooncake', 'enable_hisparse': False, 'hisparse_config': None, 'enable_broadcast_mm_inputs_process': False, 'enable_prefix_mm_cache': False, 'mm_enable_dp_encoder': False, 'mm_process_config': {}, 'mm_processor_worker_num': 0, 'mm_io_worker_num': 0, 'allowed_media_domains': [], 'media_url_max_file_size_mb': 64, 'mm_preprocess_cache_size_mb': None, 'trust_mm_content_hashes': False, 'limit_mm_data_per_request': None, 'enable_mm_global_cache': False, 'image_processor_backend': 'auto', 'mm_global_cache_backend': 'mooncake', 'disable_fast_image_processor': False, 'mm_feature_transport': 'cpu', 'enable_lora': None, 'enable_lora_overlap_loading': None, 'max_lora_rank': None, 'lora_target_modules': None, 'lora_paths': None, 'max_loaded_loras': None, 'max_loras_per_batch': 8, 'lora_eviction_policy': 'lru', 'lora_backend': 'csgmv', 'max_lora_chunk_size': 16, 'experts_shared_outer_loras': None, 'lora_use_virtual_experts': False, 'lora_strict_loading': False, 'lora_drain_wait_threshold': 0.0, 'enable_two_batch_overlap': False, 'enable_single_batch_overlap': False, 'tbo_token_distribution_threshold': 0.48, 'cpu_offload_gb': 0, 'offload_group_size': -1, 'offload_num_in_group': 1, 'offload_prefetch_step': 1, 'offload_mode': 'cpu', 'enable_lmcache': False, 'lmcache_config_file': None, 'enable_flexkv': False, 'flexkv_config_file': None, 'kt_weight_path': None, 'kt_method': 'AMXINT4', 'kt_cpuinfer': None, 'kt_threadpool_count': 2, 'kt_num_gpu_experts': None, 'kt_max_deferred_experts_per_token': None, 'dllm_algorithm': None, 'dllm_algorithm_config': None, 'dllm_fdfo': True, 'disaggregation_mode': 'null', 'disaggregation_transfer_backend': 'mooncake', 'disaggregation_bootstrap_port': 8998, 'disaggregation_ib_device': None, 'disaggregation_decode_enable_radix_cache': False, 'disaggregation_decode_enable_offload_kvcache': False, 'disaggregation_decode_retraction_backup': None, 'num_reserved_decode_tokens': 512, 'disaggregation_decode_extra_slots': None, 'disaggregation_decode_polling_interval': 1, 'optimistic_prefill_attempts': 0, 'encoder_only': False, 'language_only': False, 'language_model_only': False, 'encoder_transfer_backend': 'zmq_to_scheduler', 'encoder_urls': [], 'encoder_bootstrap_port': 8997, 'encoder_register_urls': [], 'enable_adaptive_dispatch_to_encoder': False, 'enable_pdmux': False, 'pdmux_config_path': None, 'sm_group_num': 8, 'startup_weight_load_mode': 'serial', 'custom_weight_loader': [], 'weight_loader_disable_mmap': False, 'weight_loader_prefetch_checkpoints': False, 'weight_loader_prefetch_num_threads': 4, 'weight_loader_drop_cache_after_load': False, 'remote_instance_weight_loader_seed_instance_ip': None, 'remote_instance_weight_loader_seed_instance_service_port': None, 'remote_instance_weight_loader_send_weights_group_ports': None, 'remote_instance_weight_loader_backend': 'nccl', 'remote_instance_weight_loader_start_seed_via_transfer_engine': False, 'engine_info_bootstrap_port': 6789, 'modelexpress_config': None, 'download_dir': None, 'model_checksum': None, 'delete_ckpt_after_loading': False, 'decrypted_config_file': None, 'decrypted_draft_config_file': None, 'checkpoint_engine_wait_weights_before_ready': False, 'enable_prefill_delayer': False, 'prefill_delayer_max_delay_passes': 30, 'prefill_delayer_token_usage_low_watermark': None, 'prefill_delayer_forward_passes_buckets': None, 'prefill_delayer_wait_seconds_buckets': None, 'prefill_delayer_queue_min_ratio': None, 'prefill_delayer_max_delay_ms': None, 'min_free_slots_delay': None, 'enable_deterministic_inference': True, 'rl_on_policy_target': None, 'kv_canary': 'none', 'kv_canary_real_data': 'none', 'kv_canary_sweep_interval': 0, 'enable_dynamic_batch_tokenizer': False, 'dynamic_batch_tokenizer_batch_size': 32, 'dynamic_batch_tokenizer_batch_timeout': 0.002, 'enable_tokenizer_batch_encode': False, 'disable_tokenizer_batch_decode': False, 'debug_tensor_dump_output_folder': None, 'debug_tensor_dump_layers': None, 'debug_tensor_dump_input_file': None, 'enable_memory_saver': False, 'enable_weights_cpu_backup': False, 'enable_draft_weights_cpu_backup': False, 'enable_custom_logit_processor': False, 'enable_return_hidden_states': False, 'return_hidden_states_mode': None, 'enable_return_routed_experts': False, 'enable_return_indexer_topk': False, 'sampling_mask_max_tokens': 4096, 'disable_outlines_disk_cache': False, 'enable_mis': False, 'weight_cache_mode': 'off', 'weight_cache_socket': None, 'weight_cache_timeout': 1800, 'forward_hooks': None, 'msprobe_dump_config': None, 'ple_offload_embedding': None, 'ple_offload_backend': 'pinned', 'ple_offload_dir': None, 'disaggregation_enable_kv_checksum': False, 'enable_pd_role_switch': False, 'speculative_domino_candidate_pool_size': 2048, 'flashinfer_a2a_dispatch_type': None, 'cuda_graph_max_seq_len_prefill': None, 'prefill_kv_cache_dequant_dtype': 'auto', 'dsv4_attn_backend': 'auto', 'otlp_service_name': None, '_radix_eviction_policy_explicitly_set': False, 'hicache_host_memory_fraction': 0.8, 'enable_linker_mla_dedup': False, '_swa_full_tokens_ratio_explicitly_set': False, 'swa_prefix_tails': None, 'enable_encoder_swa_bounded_replay': False, 'enable_decoder_swa_bounded_replay': False, 'enable_response_store': False, 'disaggregation_decode_host_receive_threshold': 1.0, 'qsa_indexer_dtype': 'auto', 'disaggregation_decode_allocation_policy': 'early', 'speculative_boundary_reduction': 'rs+rsv', 'speculative_use_block_verification': False, 'speculative_draft_sink_size': None, 'disable_aiter_allreduce_fusion_in_prefill': False, 'disable_aiter_allreduce_fusion_in_decode': False, 'dsa_triton_union': 0, 'enable_dsa_fused_indexer': None, 'enable_cp_tp_group_sharing': False, 'enable_watermark': False, 'watermark_config': '<redacted>', 'watermark_key': '<redacted>', 'watermark_key_b': '<redacted>', 'watermark_context_window': 4, 'watermark_mixing_probability': 0.5, 'watermark_max_probability': 1.0, 'watermark_default_enabled': False, 'watermark_enforce_all': False, 'grpc_response_timeout_secs': 300}
[2026-10-17 15:40:05] Using default HuggingFace chat template with detected content format: string
[2026-10-17 15:40:09] Init parallel begin.
[2026-10-17 15:40:09] Init parallel ends. elapsed=0.25 s
[2026-10-17 15:40:10] Memory baseline taken. mem usage=0.00 GB
/home/rohit/sglang-lab/.venv/lib/python3.12/site-packages/torch/jit/_script.py:1491: FutureWarning: `torch.jit.script` is deprecated. Please switch to `torch.compile` or `torch.export`.
  warnings.warn(
/home/rohit/sglang-lab/.venv/lib/python3.12/site-packages/torch/jit/_script.py:1643: FutureWarning: `torch.jit.interface` is deprecated. Please use `torch.compile` instead.
  warnings.warn(
/home/rohit/sglang-lab/.venv/lib/python3.12/site-packages/torch/jit/_script.py:1491: FutureWarning: `torch.jit.script` is deprecated. Please switch to `torch.compile` or `torch.export`.
  warnings.warn(
[2026-10-17 15:40:11] Load weight begin. avail mem=113.44 GB
[2026-10-17 15:40:13] Found local HF snapshot for Qwen/Qwen3-30B-A3B-Instruct-2507 at /home/rohit/sglang-lab/hf/hub/models--Qwen--Qwen3-30B-A3B-Instruct-2507/snapshots/0d7cf23991f47feeb3a57ecb4c9cee8ea4a17bfe; skipping download.

Multi-thread loading shards:   0% Completed | 0/16 [00:00<?, ?it/s]
Multi-thread loading shards:   6% Completed | 1/16 [00:17<04:21, 17.46s/it]
Multi-thread loading shards:  12% Completed | 2/16 [00:38<04:33, 19.57s/it]
Multi-thread loading shards:  19% Completed | 3/16 [00:59<04:24, 20.37s/it]
Multi-thread loading shards:  25% Completed | 4/16 [01:21<04:09, 20.78s/it]
Multi-thread loading shards:  31% Completed | 5/16 [01:42<03:49, 20.90s/it]
Multi-thread loading shards:  38% Completed | 6/16 [02:03<03:30, 21.00s/it]
Multi-thread loading shards:  44% Completed | 7/16 [02:24<03:09, 21.03s/it]
Multi-thread loading shards:  50% Completed | 8/16 [02:45<02:48, 21.04s/it]
Multi-thread loading shards:  56% Completed | 9/16 [03:06<02:26, 20.93s/it]
Multi-thread loading shards:  62% Completed | 10/16 [03:27<02:04, 20.83s/it]
Multi-thread loading shards:  69% Completed | 11/16 [03:47<01:43, 20.68s/it]
Multi-thread loading shards:  75% Completed | 12/16 [04:07<01:22, 20.55s/it]
Multi-thread loading shards:  81% Completed | 13/16 [04:27<01:01, 20.49s/it]
Multi-thread loading shards:  88% Completed | 14/16 [04:49<00:41, 20.67s/it]
Multi-thread loading shards:  94% Completed | 15/16 [05:09<00:20, 20.62s/it]
Multi-thread loading shards: 100% Completed | 16/16 [05:12<00:00, 15.35s/it]
Multi-thread loading shards: 100% Completed | 16/16 [05:12<00:00, 19.54s/it]
[2026-10-17 15:45:26] Load weight end. elapsed=315.12 s, type=Qwen3MoeForCausalLM, avail mem=55.54 GB, mem usage=57.90 GB.
[2026-10-17 15:45:29] KV Cache is allocated. dtype: torch.bfloat16, #tokens: 234517, K size: 10.74 GB, V size: 10.74 GB
[2026-10-17 15:45:29] Memory pool end. avail mem=31.45 GB
[2026-10-17 15:45:30] Capture target prefill CUDA graph begin. backend=breakable, num_tokens=[4, 8, 12, 16, 20, 24, 28, 32, 48, 64, 80, 96, 112, 128, 144, 160, 176, 192, 208, 224, 240, 256, 288, 320, 352, 384, 416, 448, 480, 512, 576, 640, 704, 768, 832, 896, 960, 1024, 1280, 1536, 1792, 2048, 2304, 2560, 2816, 3072, 3328, 3584, 3840, 4096, 4608, 5120, 5632, 6144, 6656, 7168, 7680, 8192], avail mem=29.58 GB

  0%|          | 0/58 [00:00<?, ?it/s]
Capturing prefill shape (num_tokens=8192 avail_mem=29.57 GB):   0%|          | 0/58 [00:00<?, ?it/s][2026-10-17 15:45:31] Deterministic inference is enabled, using default MoE kernel config.
[2026-10-17 15:45:31] Deterministic inference is enabled, using default MoE kernel config.

Capturing prefill shape (num_tokens=8192 avail_mem=29.57 GB):   2%|▏         | 1/58 [00:04<04:00,  4.21s/it]
Capturing prefill shape (num_tokens=7680 avail_mem=27.18 GB):   2%|▏         | 1/58 [00:04<04:00,  4.21s/it]
Capturing prefill shape (num_tokens=7680 avail_mem=27.18 GB):   3%|▎         | 2/58 [00:09<04:37,  4.95s/it]
Capturing prefill shape (num_tokens=7168 avail_mem=27.16 GB):   3%|▎         | 2/58 [00:09<04:37,  4.95s/it]
Capturing prefill shape (num_tokens=7168 avail_mem=27.16 GB):   5%|▌         | 3/58 [00:14<04:30,  4.91s/it]
Capturing prefill shape (num_tokens=6656 avail_mem=27.14 GB):   5%|▌         | 3/58 [00:14<04:30,  4.91s/it]
Capturing prefill shape (num_tokens=6656 avail_mem=27.14 GB):   7%|▋         | 4/58 [00:18<04:14,  4.71s/it]
Capturing prefill shape (num_tokens=6144 avail_mem=27.11 GB):   7%|▋         | 4/58 [00:18<04:14,  4.71s/it]
Capturing prefill shape (num_tokens=6144 avail_mem=27.11 GB):   9%|▊         | 5/58 [00:22<03:54,  4.43s/it]
Capturing prefill shape (num_tokens=5632 avail_mem=27.08 GB):   9%|▊         | 5/58 [00:22<03:54,  4.43s/it]
Capturing prefill shape (num_tokens=5632 avail_mem=27.08 GB):  10%|█         | 6/58 [00:26<03:33,  4.11s/it]
Capturing prefill shape (num_tokens=5120 avail_mem=27.04 GB):  10%|█         | 6/58 [00:26<03:33,  4.11s/it]
Capturing prefill shape (num_tokens=5120 avail_mem=27.04 GB):  12%|█▏        | 7/58 [00:29<03:12,  3.78s/it]
Capturing prefill shape (num_tokens=4608 avail_mem=27.00 GB):  12%|█▏        | 7/58 [00:29<03:12,  3.78s/it]
Capturing prefill shape (num_tokens=4608 avail_mem=27.00 GB):  14%|█▍        | 8/58 [00:32<02:51,  3.43s/it]
Capturing prefill shape (num_tokens=4096 avail_mem=26.97 GB):  14%|█▍        | 8/58 [00:32<02:51,  3.43s/it]
Capturing prefill shape (num_tokens=4096 avail_mem=26.97 GB):  16%|█▌        | 9/58 [00:34<02:29,  3.04s/it]
Capturing prefill shape (num_tokens=3840 avail_mem=26.94 GB):  16%|█▌        | 9/58 [00:34<02:29,  3.04s/it]
Capturing prefill shape (num_tokens=3840 avail_mem=26.94 GB):  17%|█▋        | 10/58 [00:36<02:09,  2.70s/it]
Capturing prefill shape (num_tokens=3584 avail_mem=26.91 GB):  17%|█▋        | 10/58 [00:36<02:09,  2.70s/it]
Capturing prefill shape (num_tokens=3584 avail_mem=26.91 GB):  19%|█▉        | 11/58 [00:38<01:53,  2.42s/it]
Capturing prefill shape (num_tokens=3328 avail_mem=26.88 GB):  19%|█▉        | 11/58 [00:38<01:53,  2.42s/it]
Capturing prefill shape (num_tokens=3328 avail_mem=26.88 GB):  21%|██        | 12/58 [00:39<01:40,  2.17s/it]
Capturing prefill shape (num_tokens=3072 avail_mem=26.85 GB):  21%|██        | 12/58 [00:39<01:40,  2.17s/it]
Capturing prefill shape (num_tokens=3072 avail_mem=26.85 GB):  22%|██▏       | 13/58 [00:41<01:28,  1.96s/it]
Capturing prefill shape (num_tokens=2816 avail_mem=26.82 GB):  22%|██▏       | 13/58 [00:41<01:28,  1.96s/it]
Capturing prefill shape (num_tokens=2816 avail_mem=26.82 GB):  24%|██▍       | 14/58 [00:42<01:17,  1.77s/it]
Capturing prefill shape (num_tokens=2560 avail_mem=26.80 GB):  24%|██▍       | 14/58 [00:42<01:17,  1.77s/it]
Capturing prefill shape (num_tokens=2560 avail_mem=26.80 GB):  26%|██▌       | 15/58 [00:43<01:07,  1.58s/it]
Capturing prefill shape (num_tokens=2304 avail_mem=26.76 GB):  26%|██▌       | 15/58 [00:43<01:07,  1.58s/it]
Capturing prefill shape (num_tokens=2304 avail_mem=26.76 GB):  28%|██▊       | 16/58 [00:44<00:58,  1.40s/it]
Capturing prefill shape (num_tokens=2048 avail_mem=26.74 GB):  28%|██▊       | 16/58 [00:44<00:58,  1.40s/it]
Capturing prefill shape (num_tokens=2048 avail_mem=26.74 GB):  29%|██▉       | 17/58 [00:45<00:50,  1.24s/it]
Capturing prefill shape (num_tokens=1792 avail_mem=26.71 GB):  29%|██▉       | 17/58 [00:45<00:50,  1.24s/it]
Capturing prefill shape (num_tokens=1792 avail_mem=26.71 GB):  31%|███       | 18/58 [00:46<00:43,  1.08s/it]
Capturing prefill shape (num_tokens=1536 avail_mem=26.68 GB):  31%|███       | 18/58 [00:46<00:43,  1.08s/it]
Capturing prefill shape (num_tokens=1536 avail_mem=26.68 GB):  33%|███▎      | 19/58 [00:46<00:36,  1.07it/s]
Capturing prefill shape (num_tokens=1280 avail_mem=26.65 GB):  33%|███▎      | 19/58 [00:46<00:36,  1.07it/s]
Capturing prefill shape (num_tokens=1280 avail_mem=26.65 GB):  34%|███▍      | 20/58 [00:47<00:30,  1.26it/s]
Capturing prefill shape (num_tokens=1024 avail_mem=26.62 GB):  34%|███▍      | 20/58 [00:47<00:30,  1.26it/s]
Capturing prefill shape (num_tokens=1024 avail_mem=26.62 GB):  36%|███▌      | 21/58 [00:47<00:24,  1.50it/s]
Capturing prefill shape (num_tokens=960 avail_mem=26.59 GB):  36%|███▌      | 21/58 [00:47<00:24,  1.50it/s] 
Capturing prefill shape (num_tokens=960 avail_mem=26.59 GB):  38%|███▊      | 22/58 [00:47<00:20,  1.80it/s]
Capturing prefill shape (num_tokens=896 avail_mem=26.56 GB):  38%|███▊      | 22/58 [00:47<00:20,  1.80it/s]
Capturing prefill shape (num_tokens=896 avail_mem=26.56 GB):  40%|███▉      | 23/58 [00:48<00:16,  2.10it/s]
Capturing prefill shape (num_tokens=832 avail_mem=26.53 GB):  40%|███▉      | 23/58 [00:48<00:16,  2.10it/s]
Capturing prefill shape (num_tokens=832 avail_mem=26.53 GB):  41%|████▏     | 24/58 [00:48<00:14,  2.40it/s]
Capturing prefill shape (num_tokens=768 avail_mem=26.50 GB):  41%|████▏     | 24/58 [00:48<00:14,  2.40it/s]
Capturing prefill shape (num_tokens=768 avail_mem=26.50 GB):  43%|████▎     | 25/58 [00:48<00:12,  2.74it/s]
Capturing prefill shape (num_tokens=704 avail_mem=26.48 GB):  43%|████▎     | 25/58 [00:48<00:12,  2.74it/s]
Capturing prefill shape (num_tokens=704 avail_mem=26.48 GB):  45%|████▍     | 26/58 [00:48<00:10,  3.09it/s]
Capturing prefill shape (num_tokens=640 avail_mem=26.44 GB):  45%|████▍     | 26/58 [00:48<00:10,  3.09it/s]
Capturing prefill shape (num_tokens=640 avail_mem=26.44 GB):  47%|████▋     | 27/58 [00:49<00:09,  3.44it/s]
Capturing prefill shape (num_tokens=576 avail_mem=26.42 GB):  47%|████▋     | 27/58 [00:49<00:09,  3.44it/s]
Capturing prefill shape (num_tokens=576 avail_mem=26.42 GB):  48%|████▊     | 28/58 [00:49<00:07,  3.85it/s]
Capturing prefill shape (num_tokens=512 avail_mem=26.39 GB):  48%|████▊     | 28/58 [00:49<00:07,  3.85it/s]
Capturing prefill shape (num_tokens=512 avail_mem=26.39 GB):  50%|█████     | 29/58 [00:49<00:06,  4.26it/s]
Capturing prefill shape (num_tokens=480 avail_mem=26.36 GB):  50%|█████     | 29/58 [00:49<00:06,  4.26it/s]
Capturing prefill shape (num_tokens=480 avail_mem=26.36 GB):  52%|█████▏    | 30/58 [00:49<00:06,  4.63it/s]
Capturing prefill shape (num_tokens=448 avail_mem=26.33 GB):  52%|█████▏    | 30/58 [00:49<00:06,  4.63it/s]
Capturing prefill shape (num_tokens=448 avail_mem=26.33 GB):  53%|█████▎    | 31/58 [00:49<00:05,  4.67it/s]
Capturing prefill shape (num_tokens=416 avail_mem=26.27 GB):  53%|█████▎    | 31/58 [00:49<00:05,  4.67it/s]
Capturing prefill shape (num_tokens=416 avail_mem=26.27 GB):  55%|█████▌    | 32/58 [00:50<00:05,  4.57it/s]
Capturing prefill shape (num_tokens=384 avail_mem=26.18 GB):  55%|█████▌    | 32/58 [00:50<00:05,  4.57it/s]
Capturing prefill shape (num_tokens=384 avail_mem=26.18 GB):  57%|█████▋    | 33/58 [00:50<00:05,  4.68it/s]
Capturing prefill shape (num_tokens=352 avail_mem=26.09 GB):  57%|█████▋    | 33/58 [00:50<00:05,  4.68it/s]
Capturing prefill shape (num_tokens=352 avail_mem=26.09 GB):  59%|█████▊    | 34/58 [00:50<00:04,  4.86it/s]
Capturing prefill shape (num_tokens=320 avail_mem=26.06 GB):  59%|█████▊    | 34/58 [00:50<00:04,  4.86it/s]
Capturing prefill shape (num_tokens=320 avail_mem=26.06 GB):  60%|██████    | 35/58 [00:50<00:04,  5.15it/s]
Capturing prefill shape (num_tokens=288 avail_mem=26.03 GB):  60%|██████    | 35/58 [00:50<00:04,  5.15it/s]
Capturing prefill shape (num_tokens=288 avail_mem=26.03 GB):  62%|██████▏   | 36/58 [00:50<00:04,  5.47it/s]
Capturing prefill shape (num_tokens=256 avail_mem=26.00 GB):  62%|██████▏   | 36/58 [00:50<00:04,  5.47it/s]
Capturing prefill shape (num_tokens=256 avail_mem=26.00 GB):  64%|██████▍   | 37/58 [00:50<00:03,  5.65it/s]
Capturing prefill shape (num_tokens=240 avail_mem=25.96 GB):  64%|██████▍   | 37/58 [00:50<00:03,  5.65it/s]
Capturing prefill shape (num_tokens=240 avail_mem=25.96 GB):  66%|██████▌   | 38/58 [00:51<00:03,  5.90it/s]
Capturing prefill shape (num_tokens=224 avail_mem=25.93 GB):  66%|██████▌   | 38/58 [00:51<00:03,  5.90it/s]
Capturing prefill shape (num_tokens=224 avail_mem=25.93 GB):  67%|██████▋   | 39/58 [00:51<00:03,  6.12it/s]
Capturing prefill shape (num_tokens=208 avail_mem=25.90 GB):  67%|██████▋   | 39/58 [00:51<00:03,  6.12it/s]
Capturing prefill shape (num_tokens=208 avail_mem=25.90 GB):  69%|██████▉   | 40/58 [00:51<00:02,  6.24it/s]
Capturing prefill shape (num_tokens=192 avail_mem=25.88 GB):  69%|██████▉   | 40/58 [00:51<00:02,  6.24it/s]
Capturing prefill shape (num_tokens=192 avail_mem=25.88 GB):  71%|███████   | 41/58 [00:51<00:02,  6.37it/s]
Capturing prefill shape (num_tokens=176 avail_mem=25.85 GB):  71%|███████   | 41/58 [00:51<00:02,  6.37it/s]
Capturing prefill shape (num_tokens=176 avail_mem=25.85 GB):  72%|███████▏  | 42/58 [00:51<00:02,  6.55it/s]
Capturing prefill shape (num_tokens=160 avail_mem=25.82 GB):  72%|███████▏  | 42/58 [00:51<00:02,  6.55it/s]
Capturing prefill shape (num_tokens=160 avail_mem=25.82 GB):  74%|███████▍  | 43/58 [00:51<00:02,  6.89it/s]
Capturing prefill shape (num_tokens=144 avail_mem=25.80 GB):  74%|███████▍  | 43/58 [00:51<00:02,  6.89it/s]
Capturing prefill shape (num_tokens=144 avail_mem=25.80 GB):  76%|███████▌  | 44/58 [00:52<00:02,  6.89it/s]
Capturing prefill shape (num_tokens=128 avail_mem=25.77 GB):  76%|███████▌  | 44/58 [00:52<00:02,  6.89it/s]
Capturing prefill shape (num_tokens=128 avail_mem=25.77 GB):  78%|███████▊  | 45/58 [00:52<00:01,  6.75it/s]
Capturing prefill shape (num_tokens=112 avail_mem=25.74 GB):  78%|███████▊  | 45/58 [00:52<00:01,  6.75it/s]
Capturing prefill shape (num_tokens=112 avail_mem=25.74 GB):  79%|███████▉  | 46/58 [00:52<00:01,  6.82it/s]
Capturing prefill shape (num_tokens=96 avail_mem=25.72 GB):  79%|███████▉  | 46/58 [00:52<00:01,  6.82it/s] 
Capturing prefill shape (num_tokens=96 avail_mem=25.72 GB):  81%|████████  | 47/58 [00:52<00:01,  6.88it/s]
Capturing prefill shape (num_tokens=80 avail_mem=25.69 GB):  81%|████████  | 47/58 [00:52<00:01,  6.88it/s]
Capturing prefill shape (num_tokens=80 avail_mem=25.69 GB):  83%|████████▎ | 48/58 [00:52<00:01,  7.11it/s]
Capturing prefill shape (num_tokens=64 avail_mem=25.66 GB):  83%|████████▎ | 48/58 [00:52<00:01,  7.11it/s]
Capturing prefill shape (num_tokens=64 avail_mem=25.66 GB):  84%|████████▍ | 49/58 [00:52<00:01,  7.07it/s]
Capturing prefill shape (num_tokens=48 avail_mem=25.63 GB):  84%|████████▍ | 49/58 [00:52<00:01,  7.07it/s]
Capturing prefill shape (num_tokens=48 avail_mem=25.63 GB):  86%|████████▌ | 50/58 [00:52<00:01,  7.29it/s]
Capturing prefill shape (num_tokens=32 avail_mem=25.60 GB):  86%|████████▌ | 50/58 [00:52<00:01,  7.29it/s]
Capturing prefill shape (num_tokens=32 avail_mem=25.60 GB):  88%|████████▊ | 51/58 [00:52<00:00,  7.37it/s]
Capturing prefill shape (num_tokens=28 avail_mem=25.58 GB):  88%|████████▊ | 51/58 [00:52<00:00,  7.37it/s]
Capturing prefill shape (num_tokens=28 avail_mem=25.58 GB):  90%|████████▉ | 52/58 [00:53<00:00,  7.32it/s]
Capturing prefill shape (num_tokens=24 avail_mem=25.69 GB):  90%|████████▉ | 52/58 [00:53<00:00,  7.32it/s]
Capturing prefill shape (num_tokens=24 avail_mem=25.69 GB):  91%|█████████▏| 53/58 [00:53<00:00,  7.18it/s]
Capturing prefill shape (num_tokens=20 avail_mem=25.67 GB):  91%|█████████▏| 53/58 [00:53<00:00,  7.18it/s]
Capturing prefill shape (num_tokens=20 avail_mem=25.67 GB):  93%|█████████▎| 54/58 [00:53<00:00,  6.83it/s]
Capturing prefill shape (num_tokens=16 avail_mem=25.63 GB):  93%|█████████▎| 54/58 [00:53<00:00,  6.83it/s]
Capturing prefill shape (num_tokens=16 avail_mem=25.63 GB):  95%|█████████▍| 55/58 [00:53<00:00,  6.16it/s]
Capturing prefill shape (num_tokens=12 avail_mem=25.54 GB):  95%|█████████▍| 55/58 [00:53<00:00,  6.16it/s]
Capturing prefill shape (num_tokens=12 avail_mem=25.54 GB):  97%|█████████▋| 56/58 [00:53<00:00,  6.56it/s]
Capturing prefill shape (num_tokens=8 avail_mem=25.49 GB):  97%|█████████▋| 56/58 [00:53<00:00,  6.56it/s] 
Capturing prefill shape (num_tokens=8 avail_mem=25.49 GB):  98%|█████████▊| 57/58 [00:53<00:00,  6.85it/s]
Capturing prefill shape (num_tokens=4 avail_mem=25.45 GB):  98%|█████████▊| 57/58 [00:53<00:00,  6.85it/s]
Capturing prefill shape (num_tokens=4 avail_mem=25.45 GB): 100%|██████████| 58/58 [00:54<00:00,  6.85it/s]
Capturing prefill shape (num_tokens=4 avail_mem=25.45 GB): 100%|██████████| 58/58 [00:54<00:00,  1.07it/s]
[2026-10-17 15:46:24] Capture target prefill CUDA graph end. elapsed=54.51 s, mem usage=3.02 GB, avail mem=26.56 GB.
[2026-10-17 15:46:24] Capture target decode CUDA graph begin. backend=full, num_tokens_per_req=1, avail mem=26.56 GB

  0%|          | 0/36 [00:00<?, ?it/s]
Capturing batches (bs=256 avail_mem=26.24 GB):   0%|          | 0/36 [00:00<?, ?it/s]
Capturing batches (bs=256 avail_mem=26.24 GB):   3%|▎         | 1/36 [00:00<00:05,  6.21it/s]
Capturing batches (bs=248 avail_mem=26.20 GB):   3%|▎         | 1/36 [00:00<00:05,  6.21it/s]
Capturing batches (bs=248 avail_mem=26.20 GB):   6%|▌         | 2/36 [00:00<00:05,  6.20it/s]
Capturing batches (bs=240 avail_mem=26.17 GB):   6%|▌         | 2/36 [00:00<00:05,  6.20it/s]
Capturing batches (bs=240 avail_mem=26.17 GB):   8%|▊         | 3/36 [00:00<00:05,  5.59it/s]
Capturing batches (bs=232 avail_mem=26.14 GB):   8%|▊         | 3/36 [00:00<00:05,  5.59it/s]
Capturing batches (bs=232 avail_mem=26.14 GB):  11%|█         | 4/36 [00:00<00:05,  5.61it/s]
Capturing batches (bs=224 avail_mem=26.11 GB):  11%|█         | 4/36 [00:00<00:05,  5.61it/s]
Capturing batches (bs=224 avail_mem=26.11 GB):  14%|█▍        | 5/36 [00:00<00:05,  5.85it/s]
Capturing batches (bs=216 avail_mem=26.06 GB):  14%|█▍        | 5/36 [00:00<00:05,  5.85it/s]
Capturing batches (bs=216 avail_mem=26.06 GB):  17%|█▋        | 6/36 [00:01<00:04,  6.04it/s]
Capturing batches (bs=208 avail_mem=26.04 GB):  17%|█▋        | 6/36 [00:01<00:04,  6.04it/s]
Capturing batches (bs=208 avail_mem=26.04 GB):  19%|█▉        | 7/36 [00:01<00:04,  6.20it/s]
Capturing batches (bs=200 avail_mem=25.99 GB):  19%|█▉        | 7/36 [00:01<00:04,  6.20it/s]
Capturing batches (bs=200 avail_mem=25.99 GB):  22%|██▏       | 8/36 [00:01<00:04,  6.32it/s]
Capturing batches (bs=192 avail_mem=25.97 GB):  22%|██▏       | 8/36 [00:01<00:04,  6.32it/s]
Capturing batches (bs=192 avail_mem=25.97 GB):  25%|██▌       | 9/36 [00:01<00:04,  6.41it/s]
Capturing batches (bs=184 avail_mem=25.92 GB):  25%|██▌       | 9/36 [00:01<00:04,  6.41it/s]
Capturing batches (bs=184 avail_mem=25.92 GB):  28%|██▊       | 10/36 [00:01<00:03,  6.52it/s]
Capturing batches (bs=176 avail_mem=25.90 GB):  28%|██▊       | 10/36 [00:01<00:03,  6.52it/s]
Capturing batches (bs=176 avail_mem=25.90 GB):  31%|███       | 11/36 [00:01<00:03,  6.58it/s]
Capturing batches (bs=168 avail_mem=25.86 GB):  31%|███       | 11/36 [00:01<00:03,  6.58it/s]
Capturing batches (bs=168 avail_mem=25.86 GB):  33%|███▎      | 12/36 [00:01<00:03,  6.65it/s]
Capturing batches (bs=160 avail_mem=25.83 GB):  33%|███▎      | 12/36 [00:01<00:03,  6.65it/s]
Capturing batches (bs=160 avail_mem=25.83 GB):  36%|███▌      | 13/36 [00:02<00:03,  6.73it/s]
Capturing batches (bs=152 avail_mem=25.79 GB):  36%|███▌      | 13/36 [00:02<00:03,  6.73it/s]
Capturing batches (bs=152 avail_mem=25.79 GB):  39%|███▉      | 14/36 [00:02<00:03,  6.77it/s]
Capturing batches (bs=144 avail_mem=25.77 GB):  39%|███▉      | 14/36 [00:02<00:03,  6.77it/s]
Capturing batches (bs=144 avail_mem=25.77 GB):  42%|████▏     | 15/36 [00:02<00:03,  6.85it/s]
Capturing batches (bs=136 avail_mem=25.72 GB):  42%|████▏     | 15/36 [00:02<00:03,  6.85it/s]
Capturing batches (bs=136 avail_mem=25.72 GB):  44%|████▍     | 16/36 [00:02<00:02,  6.90it/s]
Capturing batches (bs=128 avail_mem=25.70 GB):  44%|████▍     | 16/36 [00:02<00:02,  6.90it/s]
Capturing batches (bs=128 avail_mem=25.70 GB):  47%|████▋     | 17/36 [00:02<00:02,  7.09it/s]
Capturing batches (bs=120 avail_mem=25.66 GB):  47%|████▋     | 17/36 [00:02<00:02,  7.09it/s]
Capturing batches (bs=120 avail_mem=25.66 GB):  50%|█████     | 18/36 [00:02<00:02,  7.19it/s]
Capturing batches (bs=112 avail_mem=25.63 GB):  50%|█████     | 18/36 [00:02<00:02,  7.19it/s]
Capturing batches (bs=112 avail_mem=25.63 GB):  53%|█████▎    | 19/36 [00:02<00:02,  7.26it/s]
Capturing batches (bs=104 avail_mem=25.59 GB):  53%|█████▎    | 19/36 [00:02<00:02,  7.26it/s]
Capturing batches (bs=104 avail_mem=25.59 GB):  56%|█████▌    | 20/36 [00:03<00:02,  7.33it/s]
Capturing batches (bs=96 avail_mem=25.56 GB):  56%|█████▌    | 20/36 [00:03<00:02,  7.33it/s] 
Capturing batches (bs=96 avail_mem=25.56 GB):  58%|█████▊    | 21/36 [00:03<00:02,  7.41it/s]
Capturing batches (bs=88 avail_mem=25.52 GB):  58%|█████▊    | 21/36 [00:03<00:02,  7.41it/s]
Capturing batches (bs=88 avail_mem=25.52 GB):  61%|██████    | 22/36 [00:03<00:01,  7.46it/s]
Capturing batches (bs=80 avail_mem=25.49 GB):  61%|██████    | 22/36 [00:03<00:01,  7.46it/s]
Capturing batches (bs=80 avail_mem=25.49 GB):  64%|██████▍   | 23/36 [00:03<00:01,  7.54it/s]
Capturing batches (bs=72 avail_mem=25.45 GB):  64%|██████▍   | 23/36 [00:03<00:01,  7.54it/s]
Capturing batches (bs=72 avail_mem=25.45 GB):  67%|██████▋   | 24/36 [00:03<00:01,  7.65it/s]
Capturing batches (bs=64 avail_mem=25.42 GB):  67%|██████▋   | 24/36 [00:03<00:01,  7.65it/s]
Capturing batches (bs=64 avail_mem=25.42 GB):  69%|██████▉   | 25/36 [00:03<00:01,  7.77it/s]
Capturing batches (bs=56 avail_mem=25.38 GB):  69%|██████▉   | 25/36 [00:03<00:01,  7.77it/s]
Capturing batches (bs=56 avail_mem=25.38 GB):  72%|███████▏  | 26/36 [00:03<00:01,  7.90it/s]
Capturing batches (bs=48 avail_mem=25.36 GB):  72%|███████▏  | 26/36 [00:03<00:01,  7.90it/s]
Capturing batches (bs=48 avail_mem=25.36 GB):  75%|███████▌  | 27/36 [00:03<00:01,  8.00it/s]
Capturing batches (bs=40 avail_mem=25.31 GB):  75%|███████▌  | 27/36 [00:03<00:01,  8.00it/s]
Capturing batches (bs=40 avail_mem=25.31 GB):  78%|███████▊  | 28/36 [00:04<00:00,  8.11it/s]
Capturing batches (bs=32 avail_mem=25.29 GB):  78%|███████▊  | 28/36 [00:04<00:00,  8.11it/s]
Capturing batches (bs=32 avail_mem=25.29 GB):  81%|████████  | 29/36 [00:04<00:00,  8.19it/s]
Capturing batches (bs=24 avail_mem=25.24 GB):  81%|████████  | 29/36 [00:04<00:00,  8.19it/s]
Capturing batches (bs=24 avail_mem=25.24 GB):  83%|████████▎ | 30/36 [00:04<00:00,  8.30it/s]
Capturing batches (bs=16 avail_mem=25.22 GB):  83%|████████▎ | 30/36 [00:04<00:00,  8.30it/s]
Capturing batches (bs=16 avail_mem=25.22 GB):  86%|████████▌ | 31/36 [00:04<00:00,  8.39it/s]
Capturing batches (bs=12 avail_mem=25.18 GB):  86%|████████▌ | 31/36 [00:04<00:00,  8.39it/s]
Capturing batches (bs=12 avail_mem=25.18 GB):  89%|████████▉ | 32/36 [00:04<00:00,  8.46it/s]
Capturing batches (bs=8 avail_mem=25.15 GB):  89%|████████▉ | 32/36 [00:04<00:00,  8.46it/s] 
Capturing batches (bs=8 avail_mem=25.15 GB):  92%|█████████▏| 33/36 [00:04<00:00,  8.55it/s]
Capturing batches (bs=4 avail_mem=25.11 GB):  92%|█████████▏| 33/36 [00:04<00:00,  8.55it/s]
Capturing batches (bs=4 avail_mem=25.11 GB):  94%|█████████▍| 34/36 [00:04<00:00,  8.61it/s]
Capturing batches (bs=2 avail_mem=25.09 GB):  94%|█████████▍| 34/36 [00:04<00:00,  8.61it/s]
Capturing batches (bs=2 avail_mem=25.09 GB):  97%|█████████▋| 35/36 [00:04<00:00,  8.36it/s]
Capturing batches (bs=1 avail_mem=25.04 GB):  97%|█████████▋| 35/36 [00:04<00:00,  8.36it/s]
Capturing batches (bs=1 avail_mem=25.04 GB): 100%|██████████| 36/36 [00:04<00:00,  8.18it/s]
Capturing batches (bs=1 avail_mem=25.04 GB): 100%|██████████| 36/36 [00:04<00:00,  7.23it/s]
[2026-10-17 15:46:30] Capture target decode CUDA graph end. bs=[1, 2, 4, 8, 12, 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], elapsed=5.46 s, mem usage=1.54 GB, avail mem=25.02 GB.
[2026-10-17 15:46:30] max_total_num_tokens=234517, chunked_prefill_size=8192, max_prefill_tokens=16384, max_running_requests=2048, context_len=262144, available_gpu_mem=25.02 GB
[2026-10-17 15:46:30] Init Unified Radix Cache. Components: (<ComponentType.FULL: 0>,). Tree Core: RustUnifiedTreeCore
[2026-10-17 15:46:30] Tree cache initialized: source=default impl=UnifiedRadixCache hybrid_swa=False hybrid_ssm=False hicache_attached=False
[2026-10-17 15:46:30] Engine startup timings (s): load_weight=315.12, kv_cache_allocation=2.05, scheduler_e2e=381.02, cuda_graph={prefill=54.51, decode=5.46, target_verify=0.00, draft_prefill=0.00, draft_decode=0.00, draft_extend=0.00}, tokenizer_e2e=385.91
[2026-10-17 15:46:30] INFO:     Started server process [138116]
[2026-10-17 15:46:30] INFO:     Waiting for application startup.
[2026-10-17 15:46:30] Using default chat sampling params from model generation config: {'temperature': 0.7, 'top_k': 20, 'top_p': 0.8}
[2026-10-17 15:46:30] INFO:     Application startup complete.
[2026-10-17 15:46:30] INFO:     Uvicorn running on http://127.0.0.1:30000 (Press CTRL+C to quit)
[2026-10-17 15:46:31] INFO:     127.0.0.1:58970 - "GET /model_info HTTP/1.1" 200 OK
[2026-10-17 15:46:31] INFO:     127.0.0.1:58980 - "GET /health HTTP/1.1" 503 Service Unavailable
[2026-10-17 15:46:31] Prefill batch, #new-seq: 1, #new-token: 6, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 3.97
[2026-10-17 15:46:32] INFO:     127.0.0.1:58996 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:46:32] INFO:     127.0.0.1:59008 - "POST /freeze_gc HTTP/1.1" 200 OK
[2026-10-17 15:46:32] The server is fired up and ready to roll!
[2026-10-17 15:46:33] Prefill batch, #new-seq: 1, #new-token: 1, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: False, input throughput (token/s): 0.52
[2026-10-17 15:46:34] INFO:     127.0.0.1:59018 - "GET /health HTTP/1.1" 200 OK
[2026-10-17 15:46:34] Prefill batch, #new-seq: 1, #new-token: 24, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 20.56
[2026-10-17 15:46:36] Decode batch, #running-req: 1, #token: 58, token usage: 0.00, cuda graph: True, gen throughput (token/s): 7.02, #queue-req: 0
[2026-10-17 15:46:37] Decode batch, #running-req: 1, #token: 98, token usage: 0.00, cuda graph: True, gen throughput (token/s): 29.91, #queue-req: 0
[2026-10-17 15:46:38] INFO:     127.0.0.1:59034 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:46:38] Prefill batch, #new-seq: 1, #new-token: 24, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 6.43
[2026-10-17 15:46:38] Decode batch, #running-req: 1, #token: 30, token usage: 0.00, cuda graph: True, gen throughput (token/s): 27.45, #queue-req: 0
[2026-10-17 15:46:40] Decode batch, #running-req: 1, #token: 70, token usage: 0.00, cuda graph: True, gen throughput (token/s): 29.95, #queue-req: 0
[2026-10-17 15:46:41] Decode batch, #running-req: 1, #token: 110, token usage: 0.00, cuda graph: True, gen throughput (token/s): 30.40, #queue-req: 0
[2026-10-17 15:46:42] INFO:     127.0.0.1:59044 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:46:42] Prefill batch, #new-seq: 1, #new-token: 24, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 6.46
[2026-10-17 15:46:42] Decode batch, #running-req: 1, #token: 42, token usage: 0.00, cuda graph: True, gen throughput (token/s): 27.33, #queue-req: 0
[2026-10-17 15:46:44] Decode batch, #running-req: 1, #token: 82, token usage: 0.00, cuda graph: True, gen throughput (token/s): 30.19, #queue-req: 0
[2026-10-17 15:46:45] Decode batch, #running-req: 1, #token: 122, token usage: 0.00, cuda graph: True, gen throughput (token/s): 30.43, #queue-req: 0
[2026-10-17 15:46:45] INFO:     127.0.0.1:51912 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:46:46] Prefill batch, #new-seq: 1, #new-token: 24, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 6.47
[2026-10-17 15:46:47] Decode batch, #running-req: 1, #token: 54, token usage: 0.00, cuda graph: True, gen throughput (token/s): 27.28, #queue-req: 0
[2026-10-17 15:46:48] Decode batch, #running-req: 1, #token: 94, token usage: 0.00, cuda graph: True, gen throughput (token/s): 30.46, #queue-req: 0
[2026-10-17 15:46:49] INFO:     127.0.0.1:51914 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:46:49] Prefill batch, #new-seq: 1, #new-token: 24, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 6.49
[2026-10-17 15:46:49] Decode batch, #running-req: 1, #token: 26, token usage: 0.00, cuda graph: True, gen throughput (token/s): 27.51, #queue-req: 0
[2026-10-17 15:46:51] Decode batch, #running-req: 1, #token: 66, token usage: 0.00, cuda graph: True, gen throughput (token/s): 29.93, #queue-req: 0
[2026-10-17 15:46:52] Decode batch, #running-req: 1, #token: 106, token usage: 0.00, cuda graph: True, gen throughput (token/s): 30.04, #queue-req: 0
[2026-10-17 15:46:53] INFO:     127.0.0.1:51926 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:46:53] Prefill batch, #new-seq: 16, #new-token: 365, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 95.51
[2026-10-17 15:46:54] Decode batch, #running-req: 16, #token: 589, token usage: 0.00, cuda graph: True, gen throughput (token/s): 105.74, #queue-req: 0
[2026-10-17 15:47:00] Decode batch, #running-req: 16, #token: 1229, token usage: 0.01, cuda graph: True, gen throughput (token/s): 110.57, #queue-req: 0
[2026-10-17 15:47:06] Decode batch, #running-req: 16, #token: 1869, token usage: 0.01, cuda graph: True, gen throughput (token/s): 111.92, #queue-req: 0
[2026-10-17 15:47:08] INFO:     127.0.0.1:42956 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:47:12] Decode batch, #running-req: 15, #token: 2351, token usage: 0.01, cuda graph: True, gen throughput (token/s): 105.89, #queue-req: 0
[2026-10-17 15:47:18] Decode batch, #running-req: 15, #token: 2755, token usage: 0.01, cuda graph: True, gen throughput (token/s): 98.98, #queue-req: 0
[2026-10-17 15:47:18] INFO:     127.0.0.1:42990 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:47:24] Decode batch, #running-req: 14, #token: 3315, token usage: 0.01, cuda graph: True, gen throughput (token/s): 90.63, #queue-req: 0
[2026-10-17 15:47:30] Decode batch, #running-req: 14, #token: 3875, token usage: 0.02, cuda graph: True, gen throughput (token/s): 88.75, #queue-req: 0
[2026-10-17 15:47:30] INFO:     127.0.0.1:42964 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:47:30] INFO:     127.0.0.1:42980 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:47:30] INFO:     127.0.0.1:42984 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:47:30] INFO:     127.0.0.1:42998 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:47:30] INFO:     127.0.0.1:43010 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:47:30] INFO:     127.0.0.1:43026 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:47:30] INFO:     127.0.0.1:43024 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:47:30] INFO:     127.0.0.1:43048 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:47:30] INFO:     127.0.0.1:43050 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:47:30] INFO:     127.0.0.1:43060 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:47:30] INFO:     127.0.0.1:43064 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:47:30] INFO:     127.0.0.1:43078 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:47:30] INFO:     127.0.0.1:43076 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:47:30] INFO:     127.0.0.1:43090 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:47:31] Prefill batch, #new-seq: 1, #new-token: 24, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 0.64
[2026-10-17 15:47:31] Prefill batch, #new-seq: 7, #new-token: 154, #cached-token: 0, token usage: 0.00, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 939.98
[2026-10-17 15:47:31] Prefill batch, #new-seq: 8, #new-token: 187, #cached-token: 0, token usage: 0.00, #running-req: 8, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 1141.17
[2026-10-17 15:47:35] Decode batch, #running-req: 16, #token: 989, token usage: 0.00, cuda graph: True, gen throughput (token/s): 119.22, #queue-req: 0
[2026-10-17 15:47:41] Decode batch, #running-req: 16, #token: 1629, token usage: 0.01, cuda graph: True, gen throughput (token/s): 113.43, #queue-req: 0
[2026-10-17 15:47:45] INFO:     127.0.0.1:53550 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:47:47] Decode batch, #running-req: 15, #token: 2126, token usage: 0.01, cuda graph: True, gen throughput (token/s): 109.52, #queue-req: 0
[2026-10-17 15:47:53] Decode batch, #running-req: 15, #token: 2726, token usage: 0.01, cuda graph: True, gen throughput (token/s): 101.55, #queue-req: 0
[2026-10-17 15:47:55] INFO:     127.0.0.1:53592 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:47:59] Decode batch, #running-req: 14, #token: 3105, token usage: 0.01, cuda graph: True, gen throughput (token/s): 93.97, #queue-req: 0
[2026-10-17 15:48:05] Decode batch, #running-req: 14, #token: 3665, token usage: 0.02, cuda graph: True, gen throughput (token/s): 86.29, #queue-req: 0
[2026-10-17 15:48:08] INFO:     127.0.0.1:53560 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:08] INFO:     127.0.0.1:53568 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:08] INFO:     127.0.0.1:53574 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:08] INFO:     127.0.0.1:53590 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:08] INFO:     127.0.0.1:53602 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:08] INFO:     127.0.0.1:53614 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:08] INFO:     127.0.0.1:53616 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:08] INFO:     127.0.0.1:53626 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:08] INFO:     127.0.0.1:53636 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:08] INFO:     127.0.0.1:53642 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:08] INFO:     127.0.0.1:53654 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:08] INFO:     127.0.0.1:53664 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:08] INFO:     127.0.0.1:53674 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:08] INFO:     127.0.0.1:53678 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:08] Prefill batch, #new-seq: 1, #new-token: 24, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 0.64
[2026-10-17 15:48:09] Prefill batch, #new-seq: 7, #new-token: 154, #cached-token: 0, token usage: 0.00, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 937.78
[2026-10-17 15:48:09] Prefill batch, #new-seq: 8, #new-token: 187, #cached-token: 0, token usage: 0.00, #running-req: 8, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 1133.89
[2026-10-17 15:48:11] Decode batch, #running-req: 16, #token: 749, token usage: 0.00, cuda graph: True, gen throughput (token/s): 107.73, #queue-req: 0
[2026-10-17 15:48:16] Decode batch, #running-req: 16, #token: 1389, token usage: 0.01, cuda graph: True, gen throughput (token/s): 117.83, #queue-req: 0
[2026-10-17 15:48:22] Decode batch, #running-req: 16, #token: 2029, token usage: 0.01, cuda graph: True, gen throughput (token/s): 112.05, #queue-req: 0
[2026-10-17 15:48:23] INFO:     127.0.0.1:57376 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:28] Decode batch, #running-req: 15, #token: 2501, token usage: 0.01, cuda graph: True, gen throughput (token/s): 102.52, #queue-req: 0
[2026-10-17 15:48:33] INFO:     127.0.0.1:57354 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:34] Decode batch, #running-req: 14, #token: 2895, token usage: 0.01, cuda graph: True, gen throughput (token/s): 97.53, #queue-req: 0
[2026-10-17 15:48:41] Decode batch, #running-req: 14, #token: 3455, token usage: 0.01, cuda graph: True, gen throughput (token/s): 85.73, #queue-req: 0
[2026-10-17 15:48:46] INFO:     127.0.0.1:57368 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:46] INFO:     127.0.0.1:57382 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:46] INFO:     127.0.0.1:57390 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:46] INFO:     127.0.0.1:57400 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:46] INFO:     127.0.0.1:57414 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:46] INFO:     127.0.0.1:57428 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:46] INFO:     127.0.0.1:57432 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:46] INFO:     127.0.0.1:57446 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:46] INFO:     127.0.0.1:57452 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:46] INFO:     127.0.0.1:57462 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:46] INFO:     127.0.0.1:57476 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:46] INFO:     127.0.0.1:57482 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:46] INFO:     127.0.0.1:57498 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:46] INFO:     127.0.0.1:57500 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:48:46] Prefill batch, #new-seq: 1, #new-token: 24, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 0.65
[2026-10-17 15:48:46] Prefill batch, #new-seq: 7, #new-token: 154, #cached-token: 0, token usage: 0.00, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 929.43
[2026-10-17 15:48:46] Prefill batch, #new-seq: 8, #new-token: 187, #cached-token: 0, token usage: 0.00, #running-req: 8, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 1133.49
[2026-10-17 15:48:47] Decode batch, #running-req: 16, #token: 509, token usage: 0.00, cuda graph: True, gen throughput (token/s): 94.29, #queue-req: 0
[2026-10-17 15:48:52] Decode batch, #running-req: 16, #token: 1149, token usage: 0.00, cuda graph: True, gen throughput (token/s): 128.38, #queue-req: 0
[2026-10-17 15:48:57] Decode batch, #running-req: 16, #token: 1789, token usage: 0.01, cuda graph: True, gen throughput (token/s): 112.45, #queue-req: 0
[2026-10-17 15:49:00] INFO:     127.0.0.1:49972 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:49:03] Decode batch, #running-req: 15, #token: 2276, token usage: 0.01, cuda graph: True, gen throughput (token/s): 107.57, #queue-req: 0
[2026-10-17 15:49:09] Decode batch, #running-req: 15, #token: 2876, token usage: 0.01, cuda graph: True, gen throughput (token/s): 99.28, #queue-req: 0
[2026-10-17 15:49:10] INFO:     127.0.0.1:49980 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:49:16] Decode batch, #running-req: 14, #token: 3245, token usage: 0.01, cuda graph: True, gen throughput (token/s): 87.79, #queue-req: 0
[2026-10-17 15:49:22] Decode batch, #running-req: 14, #token: 3805, token usage: 0.02, cuda graph: True, gen throughput (token/s): 88.56, #queue-req: 0
[2026-10-17 15:49:23] INFO:     127.0.0.1:49994 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:49:23] INFO:     127.0.0.1:49988 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:49:23] INFO:     127.0.0.1:50004 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:49:23] INFO:     127.0.0.1:50018 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:49:23] INFO:     127.0.0.1:50012 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:49:23] INFO:     127.0.0.1:50034 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:49:23] INFO:     127.0.0.1:50044 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:49:23] INFO:     127.0.0.1:50054 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:49:23] INFO:     127.0.0.1:50064 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:49:23] INFO:     127.0.0.1:50074 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:49:23] INFO:     127.0.0.1:50078 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:49:23] INFO:     127.0.0.1:50084 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:49:23] INFO:     127.0.0.1:50094 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:49:23] INFO:     127.0.0.1:50106 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:49:23] Prefill batch, #new-seq: 1, #new-token: 24, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 0.65
[2026-10-17 15:49:23] Prefill batch, #new-seq: 7, #new-token: 154, #cached-token: 0, token usage: 0.00, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 934.57
[2026-10-17 15:49:24] Prefill batch, #new-seq: 8, #new-token: 187, #cached-token: 0, token usage: 0.00, #running-req: 8, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 1131.41
[2026-10-17 15:49:27] Decode batch, #running-req: 16, #token: 909, token usage: 0.00, cuda graph: True, gen throughput (token/s): 122.57, #queue-req: 0
[2026-10-17 15:49:33] Decode batch, #running-req: 16, #token: 1549, token usage: 0.01, cuda graph: True, gen throughput (token/s): 113.83, #queue-req: 0
[2026-10-17 15:49:38] INFO:     127.0.0.1:50210 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:49:38] Decode batch, #running-req: 15, #token: 2051, token usage: 0.01, cuda graph: True, gen throughput (token/s): 110.62, #queue-req: 0
[2026-10-17 15:49:44] Decode batch, #running-req: 15, #token: 2651, token usage: 0.01, cuda graph: True, gen throughput (token/s): 101.67, #queue-req: 0
[2026-10-17 15:49:48] INFO:     127.0.0.1:50218 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:49:51] Decode batch, #running-req: 14, #token: 3035, token usage: 0.01, cuda graph: True, gen throughput (token/s): 90.46, #queue-req: 0
[2026-10-17 15:49:57] Decode batch, #running-req: 14, #token: 3595, token usage: 0.02, cuda graph: True, gen throughput (token/s): 89.19, #queue-req: 0
[2026-10-17 15:50:01] INFO:     127.0.0.1:50216 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:01] INFO:     127.0.0.1:50232 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:01] INFO:     127.0.0.1:50240 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:01] INFO:     127.0.0.1:50254 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:01] INFO:     127.0.0.1:50242 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:01] INFO:     127.0.0.1:50246 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:01] INFO:     127.0.0.1:50256 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:01] INFO:     127.0.0.1:50258 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:01] INFO:     127.0.0.1:50270 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:01] INFO:     127.0.0.1:50282 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:01] INFO:     127.0.0.1:50290 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:01] INFO:     127.0.0.1:50300 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:01] INFO:     127.0.0.1:50306 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:01] INFO:     127.0.0.1:50310 - "POST /generate HTTP/1.1" 200 OK
/home/rohit/sglang-lab/.venv/lib/python3.12/site-packages/sglang/srt/entrypoints/http_server.py:476: FastAPIDeprecationWarning: ORJSONResponse is deprecated, FastAPI now serializes data directly to JSON bytes via Pydantic when a return type or response model is set, which is faster and doesn't need a custom response class. Read more in the FastAPI docs: https://fastapi.tiangolo.com/advanced/custom-response/#orjson-or-response-model and https://fastapi.tiangolo.com/tutorial/response-model/
  return await original_handler(ORJSONRequest(request.scope, request.receive))
[2026-10-17 15:50:06] INFO:     127.0.0.1:53900 - "GET /v1/models HTTP/1.1" 200 OK
[2026-10-17 15:50:06] INFO:     127.0.0.1:53904 - "GET /v1/models HTTP/1.1" 200 OK
[2026-10-17 15:50:06] INFO:     127.0.0.1:53916 - "GET /model_info HTTP/1.1" 200 OK
[2026-10-17 15:50:10] INFO:     127.0.0.1:45152 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:11] Prefill batch, #new-seq: 1, #new-token: 512, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 10.85
[2026-10-17 15:50:11] Decode batch, #running-req: 1, #token: 531, token usage: 0.00, cuda graph: True, gen throughput (token/s): 22.79, #queue-req: 0
[2026-10-17 15:50:13] INFO:     127.0.0.1:45160 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:13] INFO:     127.0.0.1:45170 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:13] INFO:     127.0.0.1:45186 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:13] INFO:     127.0.0.1:45202 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:13] INFO:     127.0.0.1:45208 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:13] INFO:     127.0.0.1:45224 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:13] INFO:     127.0.0.1:45232 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:13] INFO:     127.0.0.1:45236 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:13] Prefill batch, #new-seq: 1, #new-token: 512, #cached-token: 0, token usage: 0.01, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 216.84
[2026-10-17 15:50:14] Prefill batch, #new-seq: 6, #new-token: 2816, #cached-token: 0, token usage: 0.02, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 4055.47
[2026-10-17 15:50:14] Prefill batch, #new-seq: 1, #new-token: 512, #cached-token: 0, token usage: 0.02, #running-req: 7, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 2812.43
[2026-10-17 15:50:17] Decode batch, #running-req: 8, #token: 4064, token usage: 0.02, cuda graph: True, gen throughput (token/s): 40.03, #queue-req: 0
[2026-10-17 15:50:22] Decode batch, #running-req: 8, #token: 4384, token usage: 0.02, cuda graph: True, gen throughput (token/s): 69.35, #queue-req: 0
[2026-10-17 15:50:26] Decode batch, #running-req: 8, #token: 4704, token usage: 0.02, cuda graph: True, gen throughput (token/s): 70.34, #queue-req: 0
[2026-10-17 15:50:29] INFO:     127.0.0.1:32946 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:29] INFO:     127.0.0.1:32956 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:29] INFO:     127.0.0.1:32962 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:29] INFO:     127.0.0.1:32974 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:29] INFO:     127.0.0.1:32986 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:29] INFO:     127.0.0.1:33002 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:29] INFO:     127.0.0.1:33014 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:29] INFO:     127.0.0.1:33016 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:29] Prefill batch, #new-seq: 1, #new-token: 497, #cached-token: 0, token usage: 0.01, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 33.59
[2026-10-17 15:50:29] Prefill batch, #new-seq: 4, #new-token: 2044, #cached-token: 0, token usage: 0.02, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 3824.56
[2026-10-17 15:50:30] Prefill batch, #new-seq: 3, #new-token: 1536, #cached-token: 0, token usage: 0.02, #running-req: 5, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 3505.59
[2026-10-17 15:50:32] Decode batch, #running-req: 8, #token: 4245, token usage: 0.02, cuda graph: True, gen throughput (token/s): 56.89, #queue-req: 0
[2026-10-17 15:50:36] Decode batch, #running-req: 8, #token: 4565, token usage: 0.02, cuda graph: True, gen throughput (token/s): 74.14, #queue-req: 0
[2026-10-17 15:50:41] Decode batch, #running-req: 8, #token: 4885, token usage: 0.02, cuda graph: True, gen throughput (token/s): 73.09, #queue-req: 0
[2026-10-17 15:50:44] INFO:     127.0.0.1:39340 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:44] INFO:     127.0.0.1:39342 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:44] INFO:     127.0.0.1:39352 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:44] INFO:     127.0.0.1:39356 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:44] INFO:     127.0.0.1:39358 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:44] INFO:     127.0.0.1:39364 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:44] INFO:     127.0.0.1:39378 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:44] INFO:     127.0.0.1:39394 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:44] Prefill batch, #new-seq: 1, #new-token: 512, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 36.48
[2026-10-17 15:50:44] Prefill batch, #new-seq: 1, #new-token: 512, #cached-token: 0, token usage: 0.02, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 1720.11
[2026-10-17 15:50:45] Prefill batch, #new-seq: 6, #new-token: 3043, #cached-token: 0, token usage: 0.02, #running-req: 2, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 4157.10
[2026-10-17 15:50:46] Decode batch, #running-req: 8, #token: 4179, token usage: 0.02, cuda graph: True, gen throughput (token/s): 55.96, #queue-req: 0
[2026-10-17 15:50:51] Decode batch, #running-req: 8, #token: 4499, token usage: 0.02, cuda graph: True, gen throughput (token/s): 72.76, #queue-req: 0
[2026-10-17 15:50:55] Decode batch, #running-req: 8, #token: 4819, token usage: 0.02, cuda graph: True, gen throughput (token/s): 73.47, #queue-req: 0
[2026-10-17 15:50:59] INFO:     127.0.0.1:44052 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:59] INFO:     127.0.0.1:44060 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:59] INFO:     127.0.0.1:44062 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:59] INFO:     127.0.0.1:44076 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:59] INFO:     127.0.0.1:44086 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:59] INFO:     127.0.0.1:44096 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:59] INFO:     127.0.0.1:44100 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:59] INFO:     127.0.0.1:44116 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:50:59] Prefill batch, #new-seq: 1, #new-token: 512, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 35.50
[2026-10-17 15:51:00] Prefill batch, #new-seq: 1, #new-token: 511, #cached-token: 0, token usage: 0.02, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 1776.63
[2026-10-17 15:51:00] Prefill batch, #new-seq: 6, #new-token: 3069, #cached-token: 0, token usage: 0.02, #running-req: 2, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 4098.99
[2026-10-17 15:51:01] Decode batch, #running-req: 8, #token: 4148, token usage: 0.02, cuda graph: True, gen throughput (token/s): 53.48, #queue-req: 0
[2026-10-17 15:51:06] Decode batch, #running-req: 8, #token: 4468, token usage: 0.02, cuda graph: True, gen throughput (token/s): 66.13, #queue-req: 0
[2026-10-17 15:51:11] Decode batch, #running-req: 8, #token: 4788, token usage: 0.02, cuda graph: True, gen throughput (token/s): 66.98, #queue-req: 0
[2026-10-17 15:51:16] Decode batch, #running-req: 8, #token: 5108, token usage: 0.02, cuda graph: True, gen throughput (token/s): 66.21, #queue-req: 0
[2026-10-17 15:51:16] INFO:     127.0.0.1:59076 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:16] INFO:     127.0.0.1:59078 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:16] INFO:     127.0.0.1:59092 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:16] INFO:     127.0.0.1:59100 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:16] INFO:     127.0.0.1:59104 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:16] INFO:     127.0.0.1:59108 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:16] INFO:     127.0.0.1:59112 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:16] INFO:     127.0.0.1:59122 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:16] Prefill batch, #new-seq: 1, #new-token: 511, #cached-token: 0, token usage: 0.01, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 32.75
[2026-10-17 15:51:16] Prefill batch, #new-seq: 4, #new-token: 2044, #cached-token: 0, token usage: 0.02, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 3770.53
[2026-10-17 15:51:17] Prefill batch, #new-seq: 3, #new-token: 1535, #cached-token: 0, token usage: 0.02, #running-req: 5, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 3339.01
[2026-10-17 15:51:22] Decode batch, #running-req: 8, #token: 4410, token usage: 0.02, cuda graph: True, gen throughput (token/s): 53.32, #queue-req: 0
[2026-10-17 15:51:26] Decode batch, #running-req: 8, #token: 4730, token usage: 0.02, cuda graph: True, gen throughput (token/s): 67.92, #queue-req: 0
[2026-10-17 15:51:31] Decode batch, #running-req: 8, #token: 5050, token usage: 0.02, cuda graph: True, gen throughput (token/s): 68.41, #queue-req: 0
[2026-10-17 15:51:32] INFO:     127.0.0.1:42816 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:32] INFO:     127.0.0.1:42818 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:32] INFO:     127.0.0.1:42820 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:32] INFO:     127.0.0.1:42822 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:32] INFO:     127.0.0.1:42830 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:32] INFO:     127.0.0.1:42840 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:32] INFO:     127.0.0.1:42856 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:32] INFO:     127.0.0.1:42858 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:32] Prefill batch, #new-seq: 1, #new-token: 512, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 33.74
[2026-10-17 15:51:32] Prefill batch, #new-seq: 1, #new-token: 512, #cached-token: 0, token usage: 0.02, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 1867.98
[2026-10-17 15:51:33] Prefill batch, #new-seq: 6, #new-token: 3288, #cached-token: 0, token usage: 0.02, #running-req: 2, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 4173.71
[2026-10-17 15:51:37] Decode batch, #running-req: 8, #token: 4576, token usage: 0.02, cuda graph: True, gen throughput (token/s): 53.22, #queue-req: 0
[2026-10-17 15:51:42] Decode batch, #running-req: 8, #token: 4896, token usage: 0.02, cuda graph: True, gen throughput (token/s): 67.99, #queue-req: 0
[2026-10-17 15:51:46] Decode batch, #running-req: 8, #token: 5216, token usage: 0.02, cuda graph: True, gen throughput (token/s): 67.47, #queue-req: 0
[2026-10-17 15:51:48] INFO:     127.0.0.1:41650 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:48] INFO:     127.0.0.1:41652 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:48] INFO:     127.0.0.1:41664 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:48] INFO:     127.0.0.1:41674 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:48] INFO:     127.0.0.1:41678 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:48] INFO:     127.0.0.1:41690 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:48] INFO:     127.0.0.1:41696 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:48] INFO:     127.0.0.1:41712 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:51:48] Prefill batch, #new-seq: 1, #new-token: 512, #cached-token: 0, token usage: 0.01, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 33.76
[2026-10-17 15:51:49] Prefill batch, #new-seq: 2, #new-token: 1024, #cached-token: 0, token usage: 0.02, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 2981.83
[2026-10-17 15:51:49] Prefill batch, #new-seq: 5, #new-token: 2555, #cached-token: 0, token usage: 0.02, #running-req: 3, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 3880.89
[2026-10-17 15:51:52] Decode batch, #running-req: 8, #token: 4299, token usage: 0.02, cuda graph: True, gen throughput (token/s): 54.90, #queue-req: 0
[2026-10-17 15:51:57] Decode batch, #running-req: 8, #token: 4619, token usage: 0.02, cuda graph: True, gen throughput (token/s): 68.45, #queue-req: 0
[2026-10-17 15:52:02] Decode batch, #running-req: 8, #token: 4939, token usage: 0.02, cuda graph: True, gen throughput (token/s): 66.53, #queue-req: 0
[2026-10-17 15:52:04] INFO:     127.0.0.1:55502 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:52:04] INFO:     127.0.0.1:55506 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:52:04] INFO:     127.0.0.1:55518 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:52:04] INFO:     127.0.0.1:55530 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:52:04] INFO:     127.0.0.1:55542 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:52:04] INFO:     127.0.0.1:55546 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:52:04] INFO:     127.0.0.1:55558 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:52:04] INFO:     127.0.0.1:55560 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:52:05] Prefill batch, #new-seq: 1, #new-token: 511, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 33.43
[2026-10-17 15:52:05] Prefill batch, #new-seq: 1, #new-token: 493, #cached-token: 0, token usage: 0.02, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 1962.18
[2026-10-17 15:52:06] Prefill batch, #new-seq: 6, #new-token: 2899, #cached-token: 0, token usage: 0.02, #running-req: 2, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 3886.09
[2026-10-17 15:52:08] Decode batch, #running-req: 8, #token: 4055, token usage: 0.02, cuda graph: True, gen throughput (token/s): 54.73, #queue-req: 0
[2026-10-17 15:52:12] Decode batch, #running-req: 8, #token: 4375, token usage: 0.02, cuda graph: True, gen throughput (token/s): 75.23, #queue-req: 0
[2026-10-17 15:52:16] Decode batch, #running-req: 8, #token: 4695, token usage: 0.02, cuda graph: True, gen throughput (token/s): 74.88, #queue-req: 0
[2026-10-17 15:52:19] INFO:     127.0.0.1:46692 - "GET /server_info HTTP/1.1" 200 OK
[2026-10-17 15:52:19] INFO:     127.0.0.1:46704 - "GET /server_info HTTP/1.1" 200 OK
[2026-10-17 15:52:21] SIGTERM received. signum=None frame=None. Draining requests and shutting down...
[2026-10-17 15:52:22] Subprocess detokenizer (pid=138204) crashed with exit code -15. Triggering SIGQUIT for cleanup...
[2026-10-17 15:52:22] SIGQUIT received. signum=None, frame=None. It usually means one child failed.
[2026-10-17 15:52:22] Sleeping 5 seconds before crash diagnostics to let GPU activity settle.
[2026-10-17 15:52:27] No live scheduler processes found; skipping py-spy and CUDA coredump.
[2026-10-17 15:52:27] kill_process_tree called: parent_pid=138116, include_parent=True, pid=138116