[2026-10-17 15:08:00] Registered linear attn model: config=GigaChat35Config, backend=GDNAttnBackend, archs=['GigaChat35ForCausalLM', 'GigaChat35ForCausalLMNextN']
[2026-10-17 15:08:02] 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 15:08:02] Attention backend not specified. Use flashinfer backend by default.
[2026-10-17 15:08:04] server_args={'model_path': 'Qwen/Qwen3-30B-A3B-Instruct-2507-FP8', 'tokenizer_path': 'Qwen/Qwen3-30B-A3B-Instruct-2507-FP8', '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': 'fp8_e4m3', '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': False, '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': 302506050, '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-FP8', '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': 'flashinfer', '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': False, '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': False, '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 15:08:04] Using default HuggingFace chat template with detected content format: string
[2026-10-17 15:08:09] Init parallel begin.
[2026-10-17 15:08:09] Init parallel ends. elapsed=0.26 s
[2026-10-17 15:08:09] 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 15:08:11] Load weight begin. avail mem=113.39 GB
[2026-10-17 15:08:11] Detected fp8 checkpoint.
[2026-10-17 15:08:12] Found local HF snapshot for Qwen/Qwen3-30B-A3B-Instruct-2507-FP8 at /home/rohit/sglang-lab/hf/hub/models--Qwen--Qwen3-30B-A3B-Instruct-2507-FP8/snapshots/5a5a776300a41aaa681dd7ff0106608ef2bc90db; skipping download.

Multi-thread loading shards:   0% Completed | 0/4 [00:00<?, ?it/s]
Multi-thread loading shards:  25% Completed | 1/4 [00:54<02:43, 54.58s/it]
Multi-thread loading shards:  50% Completed | 2/4 [01:40<01:38, 49.48s/it]
Multi-thread loading shards:  75% Completed | 3/4 [02:26<00:47, 47.91s/it]
Multi-thread loading shards: 100% Completed | 4/4 [02:30<00:00, 30.65s/it]
Multi-thread loading shards: 100% Completed | 4/4 [02:30<00:00, 37.68s/it]
[2026-10-17 15:10:43] Using FP8 KV cache but no scaling factors provided. Defaulting to scaling factors of 1.0.
[2026-10-17 15:10:43] Load weight end. elapsed=152.29 s, type=Qwen3MoeForCausalLM, quant=fp8, fmt=e4m3, avail mem=83.63 GB, mem usage=29.75 GB.
[2026-10-17 15:10:51] KV Cache is allocated. dtype: torch.float8_e4m3fn, #tokens: 1083008, K size: 24.79 GB, V size: 24.79 GB
[2026-10-17 15:10:51] Memory pool end. avail mem=30.93 GB
[2026-10-17 15:10:51] 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=30.51 GB

  0%|          | 0/58 [00:00<?, ?it/s]
Capturing prefill shape (num_tokens=8192 avail_mem=30.55 GB):   0%|          | 0/58 [00:00<?, ?it/s][2026-10-17 15:10:54] Using default MoE kernel config. Performance might be sub-optimal! Config file not found at /home/rohit/sglang-lab/.venv/lib/python3.12/site-packages/sglang/srt/layers/moe/moe_runner/triton_utils/configs/triton_3_8_0/E=128,N=768,device_name=NVIDIA_GB10,dtype=fp8_w8a8,block_shape=[128, 128].json, you can create them with https://github.com/sgl-project/sglang/tree/main/benchmark/kernels/fused_moe_triton
[2026-10-17 15:10:54] Using default MoE kernel config. Performance might be sub-optimal! Config file not found at /home/rohit/sglang-lab/.venv/lib/python3.12/site-packages/sglang/srt/layers/moe/moe_runner/triton_utils/configs/triton_3_8_0/E=128,N=768,device_name=NVIDIA_GB10,dtype=fp8_w8a8,block_shape=[128, 128]_down.json, you can create them with https://github.com/sgl-project/sglang/tree/main/benchmark/kernels/fused_moe_triton

Capturing prefill shape (num_tokens=8192 avail_mem=30.55 GB):   2%|▏         | 1/58 [00:04<03:56,  4.15s/it]
Capturing prefill shape (num_tokens=7680 avail_mem=27.98 GB):   2%|▏         | 1/58 [00:04<03:56,  4.15s/it]
Capturing prefill shape (num_tokens=7680 avail_mem=27.98 GB):   3%|▎         | 2/58 [00:06<02:51,  3.05s/it]
Capturing prefill shape (num_tokens=7168 avail_mem=27.96 GB):   3%|▎         | 2/58 [00:06<02:51,  3.05s/it]
Capturing prefill shape (num_tokens=7168 avail_mem=27.96 GB):   5%|▌         | 3/58 [00:08<02:25,  2.65s/it]
Capturing prefill shape (num_tokens=6656 avail_mem=27.94 GB):   5%|▌         | 3/58 [00:08<02:25,  2.65s/it]
Capturing prefill shape (num_tokens=6656 avail_mem=27.94 GB):   7%|▋         | 4/58 [00:10<02:09,  2.40s/it]
Capturing prefill shape (num_tokens=6144 avail_mem=27.94 GB):   7%|▋         | 4/58 [00:10<02:09,  2.40s/it]
Capturing prefill shape (num_tokens=6144 avail_mem=27.94 GB):   9%|▊         | 5/58 [00:12<01:54,  2.16s/it]
Capturing prefill shape (num_tokens=5632 avail_mem=27.91 GB):   9%|▊         | 5/58 [00:12<01:54,  2.16s/it]
Capturing prefill shape (num_tokens=5632 avail_mem=27.91 GB):  10%|█         | 6/58 [00:13<01:42,  1.97s/it]
Capturing prefill shape (num_tokens=5120 avail_mem=27.90 GB):  10%|█         | 6/58 [00:13<01:42,  1.97s/it]
Capturing prefill shape (num_tokens=5120 avail_mem=27.90 GB):  12%|█▏        | 7/58 [00:15<01:32,  1.80s/it]
Capturing prefill shape (num_tokens=4608 avail_mem=27.87 GB):  12%|█▏        | 7/58 [00:15<01:32,  1.80s/it]
Capturing prefill shape (num_tokens=4608 avail_mem=27.87 GB):  14%|█▍        | 8/58 [00:16<01:20,  1.62s/it]
Capturing prefill shape (num_tokens=4096 avail_mem=27.84 GB):  14%|█▍        | 8/58 [00:16<01:20,  1.62s/it]
Capturing prefill shape (num_tokens=4096 avail_mem=27.84 GB):  16%|█▌        | 9/58 [00:17<01:11,  1.45s/it]
Capturing prefill shape (num_tokens=3840 avail_mem=27.80 GB):  16%|█▌        | 9/58 [00:17<01:11,  1.45s/it]
Capturing prefill shape (num_tokens=3840 avail_mem=27.80 GB):  17%|█▋        | 10/58 [00:18<01:02,  1.30s/it]
Capturing prefill shape (num_tokens=3584 avail_mem=27.78 GB):  17%|█▋        | 10/58 [00:18<01:02,  1.30s/it]
Capturing prefill shape (num_tokens=3584 avail_mem=27.78 GB):  19%|█▉        | 11/58 [00:19<00:55,  1.18s/it]
Capturing prefill shape (num_tokens=3328 avail_mem=27.74 GB):  19%|█▉        | 11/58 [00:19<00:55,  1.18s/it]
Capturing prefill shape (num_tokens=3328 avail_mem=27.74 GB):  21%|██        | 12/58 [00:20<00:49,  1.07s/it]
Capturing prefill shape (num_tokens=3072 avail_mem=27.70 GB):  21%|██        | 12/58 [00:20<00:49,  1.07s/it]
Capturing prefill shape (num_tokens=3072 avail_mem=27.70 GB):  22%|██▏       | 13/58 [00:21<00:43,  1.03it/s]
Capturing prefill shape (num_tokens=2816 avail_mem=27.64 GB):  22%|██▏       | 13/58 [00:21<00:43,  1.03it/s]
Capturing prefill shape (num_tokens=2816 avail_mem=27.64 GB):  24%|██▍       | 14/58 [00:21<00:38,  1.13it/s]
Capturing prefill shape (num_tokens=2560 avail_mem=27.57 GB):  24%|██▍       | 14/58 [00:21<00:38,  1.13it/s]
Capturing prefill shape (num_tokens=2560 avail_mem=27.57 GB):  26%|██▌       | 15/58 [00:22<00:34,  1.26it/s]
Capturing prefill shape (num_tokens=2304 avail_mem=27.31 GB):  26%|██▌       | 15/58 [00:22<00:34,  1.26it/s]
Capturing prefill shape (num_tokens=2304 avail_mem=27.31 GB):  28%|██▊       | 16/58 [00:22<00:30,  1.40it/s]
Capturing prefill shape (num_tokens=2048 avail_mem=27.29 GB):  28%|██▊       | 16/58 [00:22<00:30,  1.40it/s]
Capturing prefill shape (num_tokens=2048 avail_mem=27.29 GB):  29%|██▉       | 17/58 [00:23<00:26,  1.56it/s]
Capturing prefill shape (num_tokens=1792 avail_mem=27.27 GB):  29%|██▉       | 17/58 [00:23<00:26,  1.56it/s]
Capturing prefill shape (num_tokens=1792 avail_mem=27.27 GB):  31%|███       | 18/58 [00:23<00:22,  1.76it/s]
Capturing prefill shape (num_tokens=1536 avail_mem=27.26 GB):  31%|███       | 18/58 [00:23<00:22,  1.76it/s]
Capturing prefill shape (num_tokens=1536 avail_mem=27.26 GB):  33%|███▎      | 19/58 [00:24<00:19,  2.01it/s]
Capturing prefill shape (num_tokens=1280 avail_mem=27.25 GB):  33%|███▎      | 19/58 [00:24<00:19,  2.01it/s]
Capturing prefill shape (num_tokens=1280 avail_mem=27.25 GB):  34%|███▍      | 20/58 [00:24<00:16,  2.30it/s]
Capturing prefill shape (num_tokens=1024 avail_mem=27.23 GB):  34%|███▍      | 20/58 [00:24<00:16,  2.30it/s]
Capturing prefill shape (num_tokens=1024 avail_mem=27.23 GB):  36%|███▌      | 21/58 [00:24<00:13,  2.70it/s]
Capturing prefill shape (num_tokens=960 avail_mem=27.22 GB):  36%|███▌      | 21/58 [00:24<00:13,  2.70it/s] 
Capturing prefill shape (num_tokens=960 avail_mem=27.22 GB):  38%|███▊      | 22/58 [00:24<00:11,  3.17it/s]
Capturing prefill shape (num_tokens=896 avail_mem=27.20 GB):  38%|███▊      | 22/58 [00:24<00:11,  3.17it/s]
Capturing prefill shape (num_tokens=896 avail_mem=27.20 GB):  40%|███▉      | 23/58 [00:25<00:09,  3.64it/s]
Capturing prefill shape (num_tokens=832 avail_mem=27.19 GB):  40%|███▉      | 23/58 [00:25<00:09,  3.64it/s]
Capturing prefill shape (num_tokens=832 avail_mem=27.19 GB):  41%|████▏     | 24/58 [00:25<00:08,  4.10it/s]
Capturing prefill shape (num_tokens=768 avail_mem=27.17 GB):  41%|████▏     | 24/58 [00:25<00:08,  4.10it/s]
Capturing prefill shape (num_tokens=768 avail_mem=27.17 GB):  43%|████▎     | 25/58 [00:25<00:07,  4.56it/s]
Capturing prefill shape (num_tokens=704 avail_mem=27.15 GB):  43%|████▎     | 25/58 [00:25<00:07,  4.56it/s]
Capturing prefill shape (num_tokens=704 avail_mem=27.15 GB):  45%|████▍     | 26/58 [00:25<00:06,  5.01it/s]
Capturing prefill shape (num_tokens=640 avail_mem=27.14 GB):  45%|████▍     | 26/58 [00:25<00:06,  5.01it/s]
Capturing prefill shape (num_tokens=640 avail_mem=27.14 GB):  47%|████▋     | 27/58 [00:25<00:05,  5.45it/s]
Capturing prefill shape (num_tokens=576 avail_mem=27.12 GB):  47%|████▋     | 27/58 [00:25<00:05,  5.45it/s]
Capturing prefill shape (num_tokens=576 avail_mem=27.12 GB):  48%|████▊     | 28/58 [00:25<00:05,  5.86it/s]
Capturing prefill shape (num_tokens=512 avail_mem=27.11 GB):  48%|████▊     | 28/58 [00:25<00:05,  5.86it/s]
Capturing prefill shape (num_tokens=512 avail_mem=27.11 GB):  50%|█████     | 29/58 [00:25<00:04,  6.11it/s]
Capturing prefill shape (num_tokens=480 avail_mem=27.55 GB):  50%|█████     | 29/58 [00:25<00:04,  6.11it/s]
Capturing prefill shape (num_tokens=480 avail_mem=27.55 GB):  52%|█████▏    | 30/58 [00:26<00:04,  6.57it/s]
Capturing prefill shape (num_tokens=448 avail_mem=27.54 GB):  52%|█████▏    | 30/58 [00:26<00:04,  6.57it/s]
Capturing prefill shape (num_tokens=448 avail_mem=27.54 GB):  53%|█████▎    | 31/58 [00:26<00:03,  6.96it/s]
Capturing prefill shape (num_tokens=416 avail_mem=27.52 GB):  53%|█████▎    | 31/58 [00:26<00:03,  6.96it/s]
Capturing prefill shape (num_tokens=416 avail_mem=27.52 GB):  55%|█████▌    | 32/58 [00:26<00:03,  7.26it/s]
Capturing prefill shape (num_tokens=384 avail_mem=27.51 GB):  55%|█████▌    | 32/58 [00:26<00:03,  7.26it/s]
Capturing prefill shape (num_tokens=384 avail_mem=27.51 GB):  57%|█████▋    | 33/58 [00:26<00:03,  7.66it/s]
Capturing prefill shape (num_tokens=352 avail_mem=27.49 GB):  57%|█████▋    | 33/58 [00:26<00:03,  7.66it/s]
Capturing prefill shape (num_tokens=352 avail_mem=27.49 GB):  59%|█████▊    | 34/58 [00:26<00:03,  7.70it/s]
Capturing prefill shape (num_tokens=320 avail_mem=27.47 GB):  59%|█████▊    | 34/58 [00:26<00:03,  7.70it/s]
Capturing prefill shape (num_tokens=320 avail_mem=27.47 GB):  60%|██████    | 35/58 [00:26<00:03,  7.61it/s]
Capturing prefill shape (num_tokens=288 avail_mem=27.26 GB):  60%|██████    | 35/58 [00:26<00:03,  7.61it/s]
Capturing prefill shape (num_tokens=288 avail_mem=27.26 GB):  62%|██████▏   | 36/58 [00:26<00:02,  7.98it/s]
Capturing prefill shape (num_tokens=256 avail_mem=27.08 GB):  62%|██████▏   | 36/58 [00:26<00:02,  7.98it/s]
Capturing prefill shape (num_tokens=256 avail_mem=27.08 GB):  64%|██████▍   | 37/58 [00:26<00:02,  8.21it/s]
Capturing prefill shape (num_tokens=240 avail_mem=26.96 GB):  64%|██████▍   | 37/58 [00:26<00:02,  8.21it/s]
Capturing prefill shape (num_tokens=240 avail_mem=26.96 GB):  66%|██████▌   | 38/58 [00:27<00:02,  8.41it/s]
Capturing prefill shape (num_tokens=224 avail_mem=26.94 GB):  66%|██████▌   | 38/58 [00:27<00:02,  8.41it/s]
Capturing prefill shape (num_tokens=224 avail_mem=26.94 GB):  67%|██████▋   | 39/58 [00:27<00:02,  8.45it/s]
Capturing prefill shape (num_tokens=208 avail_mem=26.93 GB):  67%|██████▋   | 39/58 [00:27<00:02,  8.45it/s]
Capturing prefill shape (num_tokens=208 avail_mem=26.93 GB):  69%|██████▉   | 40/58 [00:27<00:02,  8.38it/s]
Capturing prefill shape (num_tokens=192 avail_mem=26.91 GB):  69%|██████▉   | 40/58 [00:27<00:02,  8.38it/s]
Capturing prefill shape (num_tokens=192 avail_mem=26.91 GB):  71%|███████   | 41/58 [00:27<00:02,  8.10it/s]
Capturing prefill shape (num_tokens=176 avail_mem=26.89 GB):  71%|███████   | 41/58 [00:27<00:02,  8.10it/s]
Capturing prefill shape (num_tokens=176 avail_mem=26.89 GB):  72%|███████▏  | 42/58 [00:27<00:01,  8.14it/s]
Capturing prefill shape (num_tokens=160 avail_mem=26.87 GB):  72%|███████▏  | 42/58 [00:27<00:01,  8.14it/s]
Capturing prefill shape (num_tokens=160 avail_mem=26.87 GB):  74%|███████▍  | 43/58 [00:27<00:01,  8.34it/s]
Capturing prefill shape (num_tokens=144 avail_mem=26.86 GB):  74%|███████▍  | 43/58 [00:27<00:01,  8.34it/s]
Capturing prefill shape (num_tokens=144 avail_mem=26.86 GB):  76%|███████▌  | 44/58 [00:27<00:01,  8.48it/s]
Capturing prefill shape (num_tokens=128 avail_mem=26.84 GB):  76%|███████▌  | 44/58 [00:27<00:01,  8.48it/s]
Capturing prefill shape (num_tokens=128 avail_mem=26.84 GB):  78%|███████▊  | 45/58 [00:27<00:01,  8.43it/s]
Capturing prefill shape (num_tokens=112 avail_mem=26.82 GB):  78%|███████▊  | 45/58 [00:27<00:01,  8.43it/s]
Capturing prefill shape (num_tokens=112 avail_mem=26.82 GB):  79%|███████▉  | 46/58 [00:27<00:01,  8.35it/s]
Capturing prefill shape (num_tokens=96 avail_mem=26.81 GB):  79%|███████▉  | 46/58 [00:27<00:01,  8.35it/s] 
Capturing prefill shape (num_tokens=96 avail_mem=26.81 GB):  81%|████████  | 47/58 [00:28<00:01,  8.30it/s]
Capturing prefill shape (num_tokens=80 avail_mem=26.79 GB):  81%|████████  | 47/58 [00:28<00:01,  8.30it/s]
Capturing prefill shape (num_tokens=80 avail_mem=26.79 GB):  83%|████████▎ | 48/58 [00:28<00:01,  8.52it/s]
Capturing prefill shape (num_tokens=64 avail_mem=26.78 GB):  83%|████████▎ | 48/58 [00:28<00:01,  8.52it/s]
Capturing prefill shape (num_tokens=64 avail_mem=26.78 GB):  84%|████████▍ | 49/58 [00:28<00:01,  8.57it/s]
Capturing prefill shape (num_tokens=48 avail_mem=26.76 GB):  84%|████████▍ | 49/58 [00:28<00:01,  8.57it/s]
Capturing prefill shape (num_tokens=48 avail_mem=26.76 GB):  86%|████████▌ | 50/58 [00:28<00:00,  8.69it/s]
Capturing prefill shape (num_tokens=32 avail_mem=26.74 GB):  86%|████████▌ | 50/58 [00:28<00:00,  8.69it/s]
Capturing prefill shape (num_tokens=32 avail_mem=26.74 GB):  88%|████████▊ | 51/58 [00:28<00:00,  8.35it/s]
Capturing prefill shape (num_tokens=28 avail_mem=26.73 GB):  88%|████████▊ | 51/58 [00:28<00:00,  8.35it/s]
Capturing prefill shape (num_tokens=28 avail_mem=26.73 GB):  90%|████████▉ | 52/58 [00:28<00:00,  8.27it/s]
Capturing prefill shape (num_tokens=24 avail_mem=26.72 GB):  90%|████████▉ | 52/58 [00:28<00:00,  8.27it/s]
Capturing prefill shape (num_tokens=24 avail_mem=26.72 GB):  91%|█████████▏| 53/58 [00:28<00:00,  8.60it/s]
Capturing prefill shape (num_tokens=20 avail_mem=26.70 GB):  91%|█████████▏| 53/58 [00:28<00:00,  8.60it/s]
Capturing prefill shape (num_tokens=20 avail_mem=26.70 GB):  93%|█████████▎| 54/58 [00:28<00:00,  8.83it/s]
Capturing prefill shape (num_tokens=16 avail_mem=26.68 GB):  93%|█████████▎| 54/58 [00:28<00:00,  8.83it/s]
Capturing prefill shape (num_tokens=16 avail_mem=26.68 GB):  95%|█████████▍| 55/58 [00:29<00:00,  7.61it/s]
Capturing prefill shape (num_tokens=12 avail_mem=26.99 GB):  95%|█████████▍| 55/58 [00:29<00:00,  7.61it/s]
Capturing prefill shape (num_tokens=12 avail_mem=26.99 GB):  97%|█████████▋| 56/58 [00:29<00:00,  7.91it/s]
Capturing prefill shape (num_tokens=8 avail_mem=26.97 GB):  97%|█████████▋| 56/58 [00:29<00:00,  7.91it/s] 
Capturing prefill shape (num_tokens=8 avail_mem=26.97 GB):  98%|█████████▊| 57/58 [00:29<00:00,  8.01it/s]
Capturing prefill shape (num_tokens=4 avail_mem=26.95 GB):  98%|█████████▊| 57/58 [00:29<00:00,  8.01it/s]
Capturing prefill shape (num_tokens=4 avail_mem=26.95 GB): 100%|██████████| 58/58 [00:29<00:00,  8.11it/s]
Capturing prefill shape (num_tokens=4 avail_mem=26.95 GB): 100%|██████████| 58/58 [00:29<00:00,  1.97it/s]
[2026-10-17 15:11:21] Capture target prefill CUDA graph end. elapsed=29.93 s, mem usage=2.48 GB, avail mem=28.03 GB.
[2026-10-17 15:11:21] Capture target decode CUDA graph begin. backend=full, num_tokens_per_req=1, avail mem=28.03 GB

  0%|          | 0/36 [00:00<?, ?it/s]
Capturing batches (bs=256 avail_mem=27.72 GB):   0%|          | 0/36 [00:00<?, ?it/s]
Capturing batches (bs=256 avail_mem=27.72 GB):   3%|▎         | 1/36 [00:00<00:04,  7.79it/s]
Capturing batches (bs=248 avail_mem=27.68 GB):   3%|▎         | 1/36 [00:00<00:04,  7.79it/s]
Capturing batches (bs=248 avail_mem=27.68 GB):   6%|▌         | 2/36 [00:00<00:04,  8.47it/s]
Capturing batches (bs=240 avail_mem=27.65 GB):   6%|▌         | 2/36 [00:00<00:04,  8.47it/s]
Capturing batches (bs=240 avail_mem=27.65 GB):   8%|▊         | 3/36 [00:00<00:03,  8.68it/s]
Capturing batches (bs=232 avail_mem=27.62 GB):   8%|▊         | 3/36 [00:00<00:03,  8.68it/s]
Capturing batches (bs=232 avail_mem=27.62 GB):  11%|█         | 4/36 [00:00<00:03,  8.85it/s]
Capturing batches (bs=224 avail_mem=27.59 GB):  11%|█         | 4/36 [00:00<00:03,  8.85it/s]
Capturing batches (bs=224 avail_mem=27.59 GB):  14%|█▍        | 5/36 [00:00<00:03,  8.94it/s]
Capturing batches (bs=216 avail_mem=27.55 GB):  14%|█▍        | 5/36 [00:00<00:03,  8.94it/s]
Capturing batches (bs=216 avail_mem=27.55 GB):  17%|█▋        | 6/36 [00:00<00:03,  9.04it/s]
Capturing batches (bs=208 avail_mem=27.53 GB):  17%|█▋        | 6/36 [00:00<00:03,  9.04it/s]
Capturing batches (bs=208 avail_mem=27.53 GB):  19%|█▉        | 7/36 [00:00<00:03,  9.17it/s]
Capturing batches (bs=200 avail_mem=27.50 GB):  19%|█▉        | 7/36 [00:00<00:03,  9.17it/s]
Capturing batches (bs=200 avail_mem=27.50 GB):  22%|██▏       | 8/36 [00:00<00:03,  9.29it/s]
Capturing batches (bs=192 avail_mem=27.47 GB):  22%|██▏       | 8/36 [00:00<00:03,  9.29it/s]
Capturing batches (bs=192 avail_mem=27.47 GB):  25%|██▌       | 9/36 [00:00<00:02,  9.39it/s]
Capturing batches (bs=184 avail_mem=27.44 GB):  25%|██▌       | 9/36 [00:00<00:02,  9.39it/s]
Capturing batches (bs=184 avail_mem=27.44 GB):  28%|██▊       | 10/36 [00:01<00:02,  9.43it/s]
Capturing batches (bs=176 avail_mem=27.41 GB):  28%|██▊       | 10/36 [00:01<00:02,  9.43it/s]
Capturing batches (bs=176 avail_mem=27.41 GB):  31%|███       | 11/36 [00:01<00:02,  9.51it/s]
Capturing batches (bs=168 avail_mem=27.37 GB):  31%|███       | 11/36 [00:01<00:02,  9.51it/s]
Capturing batches (bs=168 avail_mem=27.37 GB):  33%|███▎      | 12/36 [00:01<00:02,  9.55it/s]
Capturing batches (bs=160 avail_mem=27.35 GB):  33%|███▎      | 12/36 [00:01<00:02,  9.55it/s]
Capturing batches (bs=160 avail_mem=27.35 GB):  36%|███▌      | 13/36 [00:01<00:02,  9.59it/s]
Capturing batches (bs=152 avail_mem=27.31 GB):  36%|███▌      | 13/36 [00:01<00:02,  9.59it/s]
Capturing batches (bs=152 avail_mem=27.31 GB):  39%|███▉      | 14/36 [00:01<00:02,  9.64it/s]
Capturing batches (bs=144 avail_mem=27.29 GB):  39%|███▉      | 14/36 [00:01<00:02,  9.64it/s]
Capturing batches (bs=144 avail_mem=27.29 GB):  42%|████▏     | 15/36 [00:01<00:02,  9.67it/s]
Capturing batches (bs=136 avail_mem=27.25 GB):  42%|████▏     | 15/36 [00:01<00:02,  9.67it/s]
Capturing batches (bs=136 avail_mem=27.25 GB):  44%|████▍     | 16/36 [00:01<00:02,  9.68it/s]
Capturing batches (bs=128 avail_mem=27.23 GB):  44%|████▍     | 16/36 [00:01<00:02,  9.68it/s]
Capturing batches (bs=128 avail_mem=27.23 GB):  47%|████▋     | 17/36 [00:01<00:01,  9.75it/s]
Capturing batches (bs=120 avail_mem=27.19 GB):  47%|████▋     | 17/36 [00:01<00:01,  9.75it/s]
Capturing batches (bs=112 avail_mem=27.17 GB):  47%|████▋     | 17/36 [00:01<00:01,  9.75it/s]
Capturing batches (bs=112 avail_mem=27.17 GB):  53%|█████▎    | 19/36 [00:02<00:01,  9.89it/s]
Capturing batches (bs=104 avail_mem=27.13 GB):  53%|█████▎    | 19/36 [00:02<00:01,  9.89it/s]
Capturing batches (bs=96 avail_mem=27.11 GB):  53%|█████▎    | 19/36 [00:02<00:01,  9.89it/s] 
Capturing batches (bs=96 avail_mem=27.11 GB):  58%|█████▊    | 21/36 [00:02<00:01,  9.93it/s]
Capturing batches (bs=88 avail_mem=27.07 GB):  58%|█████▊    | 21/36 [00:02<00:01,  9.93it/s]
Capturing batches (bs=88 avail_mem=27.07 GB):  61%|██████    | 22/36 [00:02<00:01,  9.93it/s]
Capturing batches (bs=80 avail_mem=27.05 GB):  61%|██████    | 22/36 [00:02<00:01,  9.93it/s]
Capturing batches (bs=72 avail_mem=27.01 GB):  61%|██████    | 22/36 [00:02<00:01,  9.93it/s]
Capturing batches (bs=72 avail_mem=27.01 GB):  67%|██████▋   | 24/36 [00:02<00:01,  9.96it/s]
Capturing batches (bs=64 avail_mem=26.99 GB):  67%|██████▋   | 24/36 [00:02<00:01,  9.96it/s]
Capturing batches (bs=56 avail_mem=26.95 GB):  67%|██████▋   | 24/36 [00:02<00:01,  9.96it/s]
Capturing batches (bs=56 avail_mem=26.95 GB):  72%|███████▏  | 26/36 [00:02<00:00, 10.07it/s]
Capturing batches (bs=48 avail_mem=26.93 GB):  72%|███████▏  | 26/36 [00:02<00:00, 10.07it/s]
Capturing batches (bs=40 avail_mem=26.88 GB):  72%|███████▏  | 26/36 [00:02<00:00, 10.07it/s]
Capturing batches (bs=40 avail_mem=26.88 GB):  78%|███████▊  | 28/36 [00:02<00:00, 10.18it/s]
Capturing batches (bs=32 avail_mem=26.87 GB):  78%|███████▊  | 28/36 [00:02<00:00, 10.18it/s]
Capturing batches (bs=24 avail_mem=26.83 GB):  78%|███████▊  | 28/36 [00:02<00:00, 10.18it/s]
Capturing batches (bs=24 avail_mem=26.83 GB):  83%|████████▎ | 30/36 [00:03<00:00, 10.26it/s]
Capturing batches (bs=16 avail_mem=26.81 GB):  83%|████████▎ | 30/36 [00:03<00:00, 10.26it/s]
Capturing batches (bs=12 avail_mem=26.77 GB):  83%|████████▎ | 30/36 [00:03<00:00, 10.26it/s]
Capturing batches (bs=12 avail_mem=26.77 GB):  89%|████████▉ | 32/36 [00:03<00:00, 10.46it/s]
Capturing batches (bs=8 avail_mem=26.74 GB):  89%|████████▉ | 32/36 [00:03<00:00, 10.46it/s] 
Capturing batches (bs=4 avail_mem=26.75 GB):  89%|████████▉ | 32/36 [00:03<00:00, 10.46it/s]
Capturing batches (bs=4 avail_mem=26.75 GB):  94%|█████████▍| 34/36 [00:03<00:00, 10.61it/s]
Capturing batches (bs=2 avail_mem=26.73 GB):  94%|█████████▍| 34/36 [00:03<00:00, 10.61it/s]
Capturing batches (bs=1 avail_mem=26.68 GB):  94%|█████████▍| 34/36 [00:03<00:00, 10.61it/s]
Capturing batches (bs=1 avail_mem=26.68 GB): 100%|██████████| 36/36 [00:03<00:00,  9.77it/s]
Capturing batches (bs=1 avail_mem=26.68 GB): 100%|██████████| 36/36 [00:03<00:00,  9.72it/s]
[2026-10-17 15:11:25] 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=4.18 s, mem usage=1.53 GB, avail mem=26.50 GB.
[2026-10-17 15:11:25] max_total_num_tokens=1083008, chunked_prefill_size=8192, max_prefill_tokens=16384, max_running_requests=2115, context_len=262144, available_gpu_mem=26.50 GB
[2026-10-17 15:11:25] Init Unified Radix Cache. Components: (<ComponentType.FULL: 0>,). Tree Core: RustUnifiedTreeCore
[2026-10-17 15:11:25] Tree cache initialized: source=default impl=UnifiedRadixCache hybrid_swa=False hybrid_ssm=False hicache_attached=False
[2026-10-17 15:11:26] Engine startup timings (s): load_weight=152.29, kv_cache_allocation=7.10, scheduler_e2e=196.97, cuda_graph={prefill=29.93, decode=4.18, target_verify=0.00, draft_prefill=0.00, draft_decode=0.00, draft_extend=0.00}, tokenizer_e2e=202.07
[2026-10-17 15:11:26] INFO:     Started server process [123548]
[2026-10-17 15:11:26] INFO:     Waiting for application startup.
[2026-10-17 15:11:26] Using default chat sampling params from model generation config: {'temperature': 0.7, 'top_k': 20, 'top_p': 0.8}
[2026-10-17 15:11:26] INFO:     Application startup complete.
[2026-10-17 15:11:26] INFO:     Uvicorn running on http://127.0.0.1:30000 (Press CTRL+C to quit)
[2026-10-17 15:11:27] INFO:     127.0.0.1:35722 - "GET /model_info HTTP/1.1" 200 OK
[2026-10-17 15:11:27] 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): 4.08
[2026-10-17 15:11:27] INFO:     127.0.0.1:35724 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:11:27] INFO:     127.0.0.1:35734 - "POST /freeze_gc HTTP/1.1" 200 OK
[2026-10-17 15:11:27] The server is fired up and ready to roll!
[2026-10-17 15:11:28] 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): 1.25
[2026-10-17 15:11:28] INFO:     127.0.0.1:35738 - "GET /health HTTP/1.1" 200 OK
[2026-10-17 15:11:29] Prefill batch, #new-seq: 1, #new-token: 3, #cached-token: 2, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 2.80
[2026-10-17 15:11:29] INFO:     127.0.0.1:35754 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:11:29] Prefill batch, #new-seq: 1, #new-token: 17, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 62.58
[2026-10-17 15:11:29] INFO:     127.0.0.1:35770 - "POST /v1/chat/completions 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 15:11:36] INFO:     127.0.0.1:53864 - "GET /v1/models HTTP/1.1" 200 OK
[2026-10-17 15:11:36] INFO:     127.0.0.1:53868 - "GET /v1/models HTTP/1.1" 200 OK
[2026-10-17 15:11:37] INFO:     127.0.0.1:53870 - "GET /model_info HTTP/1.1" 200 OK
[2026-10-17 15:11:41] INFO:     127.0.0.1:45096 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:11:41] 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): 41.23
[2026-10-17 15:11:42] Decode batch, #running-req: 1, #token: 532, token usage: 0.00, cuda graph: True, gen throughput (token/s): 2.44, #queue-req: 0
[2026-10-17 15:11:43] INFO:     127.0.0.1:45108 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:11:43] Prefill batch, #new-seq: 1, #new-token: 1, #cached-token: 511, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: False, input throughput (token/s): 0.62
[2026-10-17 15:11:43] Decode batch, #running-req: 1, #token: 541, token usage: 0.00, cuda graph: True, gen throughput (token/s): 22.43, #queue-req: 0
[2026-10-17 15:11:44] Decode batch, #running-req: 1, #token: 581, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.68, #queue-req: 0
[2026-10-17 15:11:45] Decode batch, #running-req: 1, #token: 621, token usage: 0.00, cuda graph: True, gen throughput (token/s): 55.80, #queue-req: 0
[2026-10-17 15:11:45] INFO:     127.0.0.1:45122 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:11:45] Prefill batch, #new-seq: 1, #new-token: 256, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 109.60
[2026-10-17 15:11:46] Decode batch, #running-req: 1, #token: 278, token usage: 0.00, cuda graph: True, gen throughput (token/s): 52.85, #queue-req: 0
[2026-10-17 15:11:46] Decode batch, #running-req: 1, #token: 318, token usage: 0.00, cuda graph: True, gen throughput (token/s): 55.28, #queue-req: 0
[2026-10-17 15:11:47] Decode batch, #running-req: 1, #token: 358, token usage: 0.00, cuda graph: True, gen throughput (token/s): 55.31, #queue-req: 0
[2026-10-17 15:11:48] INFO:     127.0.0.1:45124 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:11:48] 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): 210.22
[2026-10-17 15:11:48] Decode batch, #running-req: 1, #token: 527, token usage: 0.00, cuda graph: True, gen throughput (token/s): 46.32, #queue-req: 0
[2026-10-17 15:11:49] Decode batch, #running-req: 1, #token: 567, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.79, #queue-req: 0
[2026-10-17 15:11:49] Decode batch, #running-req: 1, #token: 607, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.81, #queue-req: 0
[2026-10-17 15:11:50] INFO:     127.0.0.1:58896 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:11:50] 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): 215.88
[2026-10-17 15:11:50] Decode batch, #running-req: 1, #token: 520, token usage: 0.00, cuda graph: True, gen throughput (token/s): 50.95, #queue-req: 0
[2026-10-17 15:11:51] Decode batch, #running-req: 1, #token: 560, token usage: 0.00, cuda graph: True, gen throughput (token/s): 55.07, #queue-req: 0
[2026-10-17 15:11:52] Decode batch, #running-req: 1, #token: 600, token usage: 0.00, cuda graph: True, gen throughput (token/s): 55.45, #queue-req: 0
[2026-10-17 15:11:52] Decode batch, #running-req: 1, #token: 0, token usage: 0.00, cuda graph: True, gen throughput (token/s): 56.69, #queue-req: 0
[2026-10-17 15:11:52] INFO:     127.0.0.1:58908 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:11:52] 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): 211.62
[2026-10-17 15:11:53] Decode batch, #running-req: 1, #token: 553, token usage: 0.00, cuda graph: True, gen throughput (token/s): 46.02, #queue-req: 0
[2026-10-17 15:11:54] Decode batch, #running-req: 1, #token: 593, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.86, #queue-req: 0
[2026-10-17 15:11:55] Decode batch, #running-req: 1, #token: 633, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.74, #queue-req: 0
[2026-10-17 15:11:55] INFO:     127.0.0.1:58918 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:11:55] Prefill batch, #new-seq: 1, #new-token: 511, #cached-token: 1, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 208.48
[2026-10-17 15:11:56] Decode batch, #running-req: 1, #token: 546, token usage: 0.00, cuda graph: True, gen throughput (token/s): 46.36, #queue-req: 0
[2026-10-17 15:11:56] Decode batch, #running-req: 1, #token: 586, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.72, #queue-req: 0
[2026-10-17 15:11:57] Decode batch, #running-req: 1, #token: 626, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.71, #queue-req: 0
[2026-10-17 15:11:57] INFO:     127.0.0.1:58926 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:11:57] 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): 206.96
[2026-10-17 15:11:58] Decode batch, #running-req: 1, #token: 539, token usage: 0.00, cuda graph: True, gen throughput (token/s): 45.23, #queue-req: 0
[2026-10-17 15:11:59] Decode batch, #running-req: 1, #token: 579, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.77, #queue-req: 0
[2026-10-17 15:11:59] Decode batch, #running-req: 1, #token: 619, token usage: 0.00, cuda graph: True, gen throughput (token/s): 56.13, #queue-req: 0
[2026-10-17 15:12:00] INFO:     127.0.0.1:59746 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:00] 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): 214.18
[2026-10-17 15:12:00] Decode batch, #running-req: 1, #token: 532, token usage: 0.00, cuda graph: True, gen throughput (token/s): 48.83, #queue-req: 0
[2026-10-17 15:12:01] Decode batch, #running-req: 1, #token: 572, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.91, #queue-req: 0
[2026-10-17 15:12:02] Decode batch, #running-req: 1, #token: 612, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.89, #queue-req: 0
[2026-10-17 15:12:02] INFO:     127.0.0.1:59758 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:02] Prefill batch, #new-seq: 1, #new-token: 496, #cached-token: 1, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 201.53
[2026-10-17 15:12:02] Decode batch, #running-req: 1, #token: 510, token usage: 0.00, cuda graph: True, gen throughput (token/s): 45.63, #queue-req: 0
[2026-10-17 15:12:03] Decode batch, #running-req: 1, #token: 550, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.76, #queue-req: 0
[2026-10-17 15:12:04] Decode batch, #running-req: 1, #token: 590, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.86, #queue-req: 0
[2026-10-17 15:12:05] INFO:     127.0.0.1:59774 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:05] 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): 206.76
[2026-10-17 15:12:05] Decode batch, #running-req: 1, #token: 517, token usage: 0.00, cuda graph: True, gen throughput (token/s): 45.31, #queue-req: 0
[2026-10-17 15:12:06] Decode batch, #running-req: 1, #token: 557, token usage: 0.00, cuda graph: True, gen throughput (token/s): 55.03, #queue-req: 0
[2026-10-17 15:12:06] Decode batch, #running-req: 1, #token: 597, token usage: 0.00, cuda graph: True, gen throughput (token/s): 56.22, #queue-req: 0
[2026-10-17 15:12:07] Decode batch, #running-req: 1, #token: 637, token usage: 0.00, cuda graph: True, gen throughput (token/s): 56.30, #queue-req: 0
[2026-10-17 15:12:07] INFO:     127.0.0.1:59788 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:07] 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): 213.94
[2026-10-17 15:12:08] Decode batch, #running-req: 1, #token: 551, token usage: 0.00, cuda graph: True, gen throughput (token/s): 47.15, #queue-req: 0
[2026-10-17 15:12:09] Decode batch, #running-req: 1, #token: 591, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.81, #queue-req: 0
[2026-10-17 15:12:09] Decode batch, #running-req: 1, #token: 631, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.77, #queue-req: 0
[2026-10-17 15:12:09] INFO:     127.0.0.1:52382 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:10] Prefill batch, #new-seq: 1, #new-token: 509, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 207.25
[2026-10-17 15:12:10] Decode batch, #running-req: 1, #token: 541, token usage: 0.00, cuda graph: True, gen throughput (token/s): 46.09, #queue-req: 0
[2026-10-17 15:12:11] Decode batch, #running-req: 1, #token: 581, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.84, #queue-req: 0
[2026-10-17 15:12:12] Decode batch, #running-req: 1, #token: 621, token usage: 0.00, cuda graph: True, gen throughput (token/s): 56.29, #queue-req: 0
[2026-10-17 15:12:12] INFO:     127.0.0.1:52398 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:12] 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): 212.76
[2026-10-17 15:12:12] Decode batch, #running-req: 1, #token: 537, token usage: 0.00, cuda graph: True, gen throughput (token/s): 47.70, #queue-req: 0
[2026-10-17 15:12:13] Decode batch, #running-req: 1, #token: 577, token usage: 0.00, cuda graph: True, gen throughput (token/s): 48.00, #queue-req: 0
[2026-10-17 15:12:14] Decode batch, #running-req: 1, #token: 617, token usage: 0.00, cuda graph: True, gen throughput (token/s): 51.92, #queue-req: 0
[2026-10-17 15:12:14] INFO:     127.0.0.1:52402 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:15] 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): 198.08
[2026-10-17 15:12:15] Decode batch, #running-req: 1, #token: 530, token usage: 0.00, cuda graph: True, gen throughput (token/s): 46.85, #queue-req: 0
[2026-10-17 15:12:16] Decode batch, #running-req: 1, #token: 570, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.74, #queue-req: 0
[2026-10-17 15:12:16] Decode batch, #running-req: 1, #token: 610, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.50, #queue-req: 0
[2026-10-17 15:12:17] INFO:     127.0.0.1:52404 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:17] 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): 207.49
[2026-10-17 15:12:17] Decode batch, #running-req: 1, #token: 523, token usage: 0.00, cuda graph: True, gen throughput (token/s): 45.79, #queue-req: 0
[2026-10-17 15:12:18] Decode batch, #running-req: 1, #token: 563, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.70, #queue-req: 0
[2026-10-17 15:12:19] Decode batch, #running-req: 1, #token: 603, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.67, #queue-req: 0
[2026-10-17 15:12:19] INFO:     127.0.0.1:52408 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:19] Prefill batch, #new-seq: 1, #new-token: 511, #cached-token: 1, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 208.76
[2026-10-17 15:12:20] Decode batch, #running-req: 1, #token: 516, token usage: 0.00, cuda graph: True, gen throughput (token/s): 46.62, #queue-req: 0
[2026-10-17 15:12:20] Decode batch, #running-req: 1, #token: 556, token usage: 0.00, cuda graph: True, gen throughput (token/s): 54.97, #queue-req: 0
[2026-10-17 15:12:21] Decode batch, #running-req: 1, #token: 596, token usage: 0.00, cuda graph: True, gen throughput (token/s): 56.43, #queue-req: 0
[2026-10-17 15:12:22] Decode batch, #running-req: 1, #token: 636, token usage: 0.00, cuda graph: True, gen throughput (token/s): 56.30, #queue-req: 0
[2026-10-17 15:12:22] INFO:     127.0.0.1:57416 - "GET /server_info HTTP/1.1" 200 OK
[2026-10-17 15:12:22] INFO:     127.0.0.1:57418 - "GET /server_info HTTP/1.1" 200 OK
[2026-10-17 15:12:30] INFO:     127.0.0.1:33870 - "GET /v1/models HTTP/1.1" 200 OK
[2026-10-17 15:12:30] INFO:     127.0.0.1:33872 - "GET /v1/models HTTP/1.1" 200 OK
[2026-10-17 15:12:31] INFO:     127.0.0.1:33874 - "GET /model_info HTTP/1.1" 200 OK
[2026-10-17 15:12:35] INFO:     127.0.0.1:33880 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:35] Prefill batch, #new-seq: 1, #new-token: 1, #cached-token: 511, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: False, input throughput (token/s): 0.07
[2026-10-17 15:12:36] INFO:     127.0.0.1:33890 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:36] INFO:     127.0.0.1:33904 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:36] INFO:     127.0.0.1:33920 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:36] INFO:     127.0.0.1:33922 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:36] INFO:     127.0.0.1:33938 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:36] INFO:     127.0.0.1:33942 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:36] INFO:     127.0.0.1:33948 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:36] INFO:     127.0.0.1:33964 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:36] Prefill batch, #new-seq: 1, #new-token: 1, #cached-token: 511, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: False, input throughput (token/s): 0.62
[2026-10-17 15:12:36] Prefill batch, #new-seq: 7, #new-token: 7, #cached-token: 3321, token usage: 0.00, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 134.77
[2026-10-17 15:12:37] Decode batch, #running-req: 8, #token: 3887, token usage: 0.00, cuda graph: True, gen throughput (token/s): 5.03, #queue-req: 0
[2026-10-17 15:12:39] Decode batch, #running-req: 8, #token: 4207, token usage: 0.00, cuda graph: True, gen throughput (token/s): 147.25, #queue-req: 0
[2026-10-17 15:12:41] Decode batch, #running-req: 8, #token: 4527, token usage: 0.00, cuda graph: True, gen throughput (token/s): 146.99, #queue-req: 0
[2026-10-17 15:12:43] Decode batch, #running-req: 8, #token: 4847, token usage: 0.00, cuda graph: True, gen throughput (token/s): 149.68, #queue-req: 0
[2026-10-17 15:12:43] INFO:     127.0.0.1:54310 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:43] INFO:     127.0.0.1:54322 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:43] INFO:     127.0.0.1:54328 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:43] INFO:     127.0.0.1:54340 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:43] INFO:     127.0.0.1:54354 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:43] INFO:     127.0.0.1:54362 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:43] INFO:     127.0.0.1:54364 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:43] INFO:     127.0.0.1:54372 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:43] Prefill batch, #new-seq: 1, #new-token: 1, #cached-token: 496, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: False, input throughput (token/s): 0.14
[2026-10-17 15:12:43] Prefill batch, #new-seq: 7, #new-token: 7, #cached-token: 3573, token usage: 0.00, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 144.00
[2026-10-17 15:12:45] Decode batch, #running-req: 8, #token: 4388, token usage: 0.00, cuda graph: True, gen throughput (token/s): 148.11, #queue-req: 0
[2026-10-17 15:12:47] Decode batch, #running-req: 8, #token: 4708, token usage: 0.00, cuda graph: True, gen throughput (token/s): 155.43, #queue-req: 0
[2026-10-17 15:12:49] Decode batch, #running-req: 8, #token: 5028, token usage: 0.00, cuda graph: True, gen throughput (token/s): 156.14, #queue-req: 0
[2026-10-17 15:12:50] INFO:     127.0.0.1:55428 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:50] INFO:     127.0.0.1:55432 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:50] INFO:     127.0.0.1:55442 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:50] INFO:     127.0.0.1:55448 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:50] INFO:     127.0.0.1:55458 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:50] INFO:     127.0.0.1:55466 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:50] INFO:     127.0.0.1:55470 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:50] INFO:     127.0.0.1:55480 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:50] Prefill batch, #new-seq: 1, #new-token: 511, #cached-token: 1, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 76.22
[2026-10-17 15:12:50] Prefill batch, #new-seq: 7, #new-token: 3553, #cached-token: 2, token usage: 0.00, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 7856.18
[2026-10-17 15:12:52] Decode batch, #running-req: 8, #token: 4323, token usage: 0.00, cuda graph: True, gen throughput (token/s): 114.97, #queue-req: 0
[2026-10-17 15:12:54] Decode batch, #running-req: 8, #token: 4643, token usage: 0.00, cuda graph: True, gen throughput (token/s): 148.68, #queue-req: 0
[2026-10-17 15:12:56] Decode batch, #running-req: 8, #token: 4963, token usage: 0.00, cuda graph: True, gen throughput (token/s): 149.60, #queue-req: 0
[2026-10-17 15:12:57] INFO:     127.0.0.1:55482 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:57] INFO:     127.0.0.1:55484 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:57] INFO:     127.0.0.1:55500 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:57] INFO:     127.0.0.1:55504 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:57] INFO:     127.0.0.1:55520 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:57] INFO:     127.0.0.1:55532 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:57] INFO:     127.0.0.1:55544 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:57] INFO:     127.0.0.1:55552 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:12:57] 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): 72.96
[2026-10-17 15:12:58] Prefill batch, #new-seq: 7, #new-token: 3572, #cached-token: 8, token usage: 0.00, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 7695.07
[2026-10-17 15:12:59] Decode batch, #running-req: 8, #token: 4292, token usage: 0.00, cuda graph: True, gen throughput (token/s): 110.73, #queue-req: 0
[2026-10-17 15:13:02] Decode batch, #running-req: 8, #token: 4612, token usage: 0.00, cuda graph: True, gen throughput (token/s): 141.01, #queue-req: 0
[2026-10-17 15:13:04] Decode batch, #running-req: 8, #token: 4932, token usage: 0.00, cuda graph: True, gen throughput (token/s): 141.49, #queue-req: 0
[2026-10-17 15:13:05] INFO:     127.0.0.1:32932 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:05] INFO:     127.0.0.1:32946 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:05] INFO:     127.0.0.1:32954 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:05] INFO:     127.0.0.1:32964 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:05] INFO:     127.0.0.1:32974 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:05] INFO:     127.0.0.1:32982 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:05] INFO:     127.0.0.1:32986 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:05] INFO:     127.0.0.1:32996 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:05] 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): 69.29
[2026-10-17 15:13:06] Prefill batch, #new-seq: 7, #new-token: 3579, #cached-token: 0, token usage: 0.00, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 8496.71
[2026-10-17 15:13:07] Decode batch, #running-req: 8, #token: 4234, token usage: 0.00, cuda graph: True, gen throughput (token/s): 114.18, #queue-req: 0
[2026-10-17 15:13:09] Decode batch, #running-req: 8, #token: 4554, token usage: 0.00, cuda graph: True, gen throughput (token/s): 141.21, #queue-req: 0
[2026-10-17 15:13:11] Decode batch, #running-req: 8, #token: 4874, token usage: 0.00, cuda graph: True, gen throughput (token/s): 142.89, #queue-req: 0
[2026-10-17 15:13:13] INFO:     127.0.0.1:56996 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:13] INFO:     127.0.0.1:57004 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:13] INFO:     127.0.0.1:57014 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:13] INFO:     127.0.0.1:57030 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:13] INFO:     127.0.0.1:57038 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:13] INFO:     127.0.0.1:57052 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:13] INFO:     127.0.0.1:57054 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:13] INFO:     127.0.0.1:57060 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:13] 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): 70.54
[2026-10-17 15:13:14] Prefill batch, #new-seq: 7, #new-token: 3797, #cached-token: 3, token usage: 0.00, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 6759.15
[2026-10-17 15:13:14] Decode batch, #running-req: 8, #token: 4400, token usage: 0.00, cuda graph: True, gen throughput (token/s): 107.63, #queue-req: 0
[2026-10-17 15:13:16] Decode batch, #running-req: 8, #token: 4720, token usage: 0.00, cuda graph: True, gen throughput (token/s): 145.66, #queue-req: 0
[2026-10-17 15:13:18] Decode batch, #running-req: 8, #token: 5040, token usage: 0.00, cuda graph: True, gen throughput (token/s): 147.57, #queue-req: 0
[2026-10-17 15:13:20] INFO:     127.0.0.1:53336 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:20] INFO:     127.0.0.1:53340 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:20] INFO:     127.0.0.1:53348 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:20] INFO:     127.0.0.1:53354 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:20] INFO:     127.0.0.1:53360 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:20] INFO:     127.0.0.1:53364 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:20] INFO:     127.0.0.1:53366 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:20] INFO:     127.0.0.1:53374 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:21] 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): 72.47
[2026-10-17 15:13:21] Prefill batch, #new-seq: 6, #new-token: 3065, #cached-token: 3, token usage: 0.00, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 7458.36
[2026-10-17 15:13:21] Prefill batch, #new-seq: 1, #new-token: 511, #cached-token: 0, token usage: 0.00, #running-req: 7, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 3676.14
[2026-10-17 15:13:21] Decode batch, #running-req: 8, #token: 4122, token usage: 0.00, cuda graph: True, gen throughput (token/s): 113.36, #queue-req: 0
[2026-10-17 15:13:23] Decode batch, #running-req: 8, #token: 4442, token usage: 0.00, cuda graph: True, gen throughput (token/s): 148.24, #queue-req: 0
[2026-10-17 15:13:26] Decode batch, #running-req: 8, #token: 4762, token usage: 0.00, cuda graph: True, gen throughput (token/s): 142.78, #queue-req: 0
[2026-10-17 15:13:28] Decode batch, #running-req: 8, #token: 5082, token usage: 0.00, cuda graph: True, gen throughput (token/s): 140.04, #queue-req: 0
[2026-10-17 15:13:28] INFO:     127.0.0.1:53376 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:28] INFO:     127.0.0.1:53386 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:28] INFO:     127.0.0.1:53390 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:28] INFO:     127.0.0.1:53400 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:28] INFO:     127.0.0.1:53416 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:28] INFO:     127.0.0.1:53426 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:28] INFO:     127.0.0.1:53442 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:28] INFO:     127.0.0.1:53456 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:28] Prefill batch, #new-seq: 1, #new-token: 510, #cached-token: 1, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 70.48
[2026-10-17 15:13:29] Prefill batch, #new-seq: 7, #new-token: 3387, #cached-token: 5, token usage: 0.00, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 7412.00
[2026-10-17 15:13:31] Decode batch, #running-req: 8, #token: 4199, token usage: 0.00, cuda graph: True, gen throughput (token/s): 120.07, #queue-req: 0
[2026-10-17 15:13:33] Decode batch, #running-req: 8, #token: 4519, token usage: 0.00, cuda graph: True, gen throughput (token/s): 156.33, #queue-req: 0
[2026-10-17 15:13:35] Decode batch, #running-req: 8, #token: 4839, token usage: 0.00, cuda graph: True, gen throughput (token/s): 158.21, #queue-req: 0
[2026-10-17 15:13:35] INFO:     127.0.0.1:35418 - "GET /server_info HTTP/1.1" 200 OK
[2026-10-17 15:13:35] INFO:     127.0.0.1:35434 - "GET /server_info HTTP/1.1" 200 OK
[2026-10-17 15:13:43] INFO:     127.0.0.1:57618 - "GET /v1/models HTTP/1.1" 200 OK
[2026-10-17 15:13:43] INFO:     127.0.0.1:57630 - "GET /v1/models HTTP/1.1" 200 OK
[2026-10-17 15:13:43] INFO:     127.0.0.1:57638 - "GET /model_info HTTP/1.1" 200 OK
[2026-10-17 15:13:47] INFO:     127.0.0.1:57642 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:47] Prefill batch, #new-seq: 1, #new-token: 1, #cached-token: 511, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: False, input throughput (token/s): 0.06
[2026-10-17 15:13:48] Decode batch, #running-req: 1, #token: 542, token usage: 0.00, cuda graph: True, gen throughput (token/s): 9.15, #queue-req: 0
[2026-10-17 15:13:49] INFO:     127.0.0.1:57648 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57656 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57658 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57672 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57686 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57702 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57712 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57716 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57722 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57730 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57746 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57748 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57750 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57754 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57766 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57782 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57788 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57804 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57818 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57832 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57836 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57850 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57856 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57868 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57872 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57888 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57896 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57912 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57928 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57944 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] Prefill batch, #new-seq: 1, #new-token: 1, #cached-token: 511, token usage: 0.01, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: False, input throughput (token/s): 0.61
[2026-10-17 15:13:49] INFO:     127.0.0.1:57948 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] INFO:     127.0.0.1:57950 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:13:49] Prefill batch, #new-seq: 20, #new-token: 20, #cached-token: 9916, token usage: 0.01, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 235.66
[2026-10-17 15:13:49] Prefill batch, #new-seq: 9, #new-token: 9, #cached-token: 4597, token usage: 0.01, #running-req: 21, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 147.07
[2026-10-17 15:13:49] Prefill batch, #new-seq: 2, #new-token: 2, #cached-token: 1020, token usage: 0.01, #running-req: 30, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 75.12
[2026-10-17 15:13:53] Decode batch, #running-req: 32, #token: 17311, token usage: 0.02, cuda graph: True, gen throughput (token/s): 237.05, #queue-req: 0
[2026-10-17 15:13:57] Decode batch, #running-req: 32, #token: 18591, token usage: 0.02, cuda graph: True, gen throughput (token/s): 320.43, #queue-req: 0
[2026-10-17 15:14:01] Decode batch, #running-req: 32, #token: 19871, token usage: 0.02, cuda graph: True, gen throughput (token/s): 317.67, #queue-req: 0
[2026-10-17 15:14:02] INFO:     127.0.0.1:44146 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44148 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44162 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44170 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44180 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44196 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44198 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44206 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44218 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44224 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44232 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44242 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44248 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44264 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44270 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44274 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44286 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44288 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44292 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44304 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44314 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44322 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44336 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] Prefill batch, #new-seq: 1, #new-token: 1, #cached-token: 510, token usage: 0.01, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: False, input throughput (token/s): 0.08
[2026-10-17 15:14:02] INFO:     127.0.0.1:44350 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44362 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44370 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44382 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44396 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44410 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44420 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44426 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] INFO:     127.0.0.1:44436 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:02] Prefill batch, #new-seq: 16, #new-token: 16, #cached-token: 8387, token usage: 0.01, #running-req: 1, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 210.58
[2026-10-17 15:14:02] Prefill batch, #new-seq: 6, #new-token: 6, #cached-token: 3062, token usage: 0.02, #running-req: 17, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 111.06
[2026-10-17 15:14:02] Prefill batch, #new-seq: 9, #new-token: 9, #cached-token: 4405, token usage: 0.02, #running-req: 23, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 169.73
[2026-10-17 15:14:05] Decode batch, #running-req: 32, #token: 17417, token usage: 0.02, cuda graph: True, gen throughput (token/s): 306.97, #queue-req: 0
[2026-10-17 15:14:09] Decode batch, #running-req: 32, #token: 18697, token usage: 0.02, cuda graph: True, gen throughput (token/s): 320.31, #queue-req: 0
[2026-10-17 15:14:13] Decode batch, #running-req: 32, #token: 19977, token usage: 0.02, cuda graph: True, gen throughput (token/s): 317.36, #queue-req: 0
[2026-10-17 15:14:15] INFO:     127.0.0.1:47100 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47112 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47114 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47118 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47132 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47148 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47150 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47162 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47178 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47188 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47200 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47202 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47214 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47216 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47222 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47238 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47244 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47246 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47254 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47264 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47268 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47280 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47282 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47292 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47300 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47304 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47316 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47332 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47342 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47350 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47362 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] INFO:     127.0.0.1:47370 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:15] Prefill batch, #new-seq: 1, #new-token: 510, #cached-token: 2, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 39.90
[2026-10-17 15:14:15] Prefill batch, #new-seq: 2, #new-token: 1022, #cached-token: 2, token usage: 0.01, #running-req: 1, #queue-req: 12, #pending-token: 0, cuda graph: True, input throughput (token/s): 4805.78
[2026-10-17 15:14:16] Prefill batch, #new-seq: 17, #new-token: 8192, #cached-token: 17, token usage: 0.02, #running-req: 3, #queue-req: 0, #pending-token: 6573, cuda graph: True, input throughput (token/s): 6840.31
[2026-10-17 15:14:17] Prefill batch, #new-seq: 13, #new-token: 6556, #cached-token: 17, token usage: 0.02, #running-req: 19, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 8232.80
[2026-10-17 15:14:19] Decode batch, #running-req: 32, #token: 17085, token usage: 0.02, cuda graph: True, gen throughput (token/s): 200.93, #queue-req: 0
[2026-10-17 15:14:23] Decode batch, #running-req: 32, #token: 18365, token usage: 0.02, cuda graph: True, gen throughput (token/s): 317.31, #queue-req: 0
[2026-10-17 15:14:27] Decode batch, #running-req: 32, #token: 19645, token usage: 0.02, cuda graph: True, gen throughput (token/s): 318.02, #queue-req: 0
[2026-10-17 15:14:30] INFO:     127.0.0.1:45538 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45548 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45554 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45558 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45574 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45590 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45592 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45598 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45612 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45622 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45630 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45638 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45652 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45662 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45664 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45672 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45680 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45694 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45704 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45708 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45710 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45726 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45732 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45738 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45742 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45752 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45762 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45770 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45776 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45784 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45788 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] INFO:     127.0.0.1:45796 - "POST /generate HTTP/1.1" 200 OK
[2026-10-17 15:14:30] Prefill batch, #new-seq: 1, #new-token: 477, #cached-token: 1, token usage: 0.00, #running-req: 0, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 36.77
[2026-10-17 15:14:30] Prefill batch, #new-seq: 8, #new-token: 4077, #cached-token: 12, token usage: 0.01, #running-req: 1, #queue-req: 6, #pending-token: 0, cuda graph: True, input throughput (token/s): 7778.51
[2026-10-17 15:14:31] Prefill batch, #new-seq: 17, #new-token: 8192, #cached-token: 25, token usage: 0.02, #running-req: 9, #queue-req: 0, #pending-token: 3555, cuda graph: True, input throughput (token/s): 8458.46
[2026-10-17 15:14:32] Prefill batch, #new-seq: 7, #new-token: 3544, #cached-token: 11, token usage: 0.02, #running-req: 25, #queue-req: 0, #pending-token: 0, cuda graph: True, input throughput (token/s): 7731.22
[2026-10-17 15:14:34] Decode batch, #running-req: 32, #token: 16889, token usage: 0.02, cuda graph: True, gen throughput (token/s): 205.66, #queue-req: 0
[2026-10-17 15:14:38] Decode batch, #running-req: 32, #token: 18169, token usage: 0.02, cuda graph: True, gen throughput (token/s): 305.01, #queue-req: 0
[2026-10-17 15:14:42] Decode batch, #running-req: 32, #token: 19449, token usage: 0.02, cuda graph: True, gen throughput (token/s): 310.04, #queue-req: 0
[2026-10-17 15:14:45] INFO:     127.0.0.1:34394 - "GET /server_info HTTP/1.1" 200 OK
[2026-10-17 15:14:45] INFO:     127.0.0.1:34406 - "GET /server_info HTTP/1.1" 200 OK
[2026-10-17 15:14:46] SIGTERM received. signum=None frame=None. Draining requests and shutting down...
[2026-10-17 15:14:47] Subprocess detokenizer (pid=123818) crashed with exit code -15. Triggering SIGQUIT for cleanup...
[2026-10-17 15:14:47] SIGQUIT received. signum=None, frame=None. It usually means one child failed.
[2026-10-17 15:14:47] Sleeping 5 seconds before crash diagnostics to let GPU activity settle.
[2026-10-17 15:14:52] No live scheduler processes found; skipping py-spy and CUDA coredump.
[2026-10-17 15:14:52] kill_process_tree called: parent_pid=123548, include_parent=True, pid=123548