https://schema.ai/vllm https://schema.ai/vllm/claims https://schema.ai/vllm/sources https://schema.ai/vllm/metrics https://schema.ai/vllm/settings https://schema.ai/vllm/surfaces https://schema.ai/vllm/topics https://schema.ai/vllm/scenarios https://schema.ai/vllm/docs https://schema.ai/vllm/docs/graph-model https://schema.ai/vllm/surfaces/entity.vllm.config.scheduler_config https://schema.ai/vllm/settings/entity.vllm.parameter.max_num_batched_tokens https://schema.ai/vllm/settings/entity.vllm.parameter.max_num_seqs https://schema.ai/vllm/settings/entity.vllm.parameter.max_model_len https://schema.ai/vllm/surfaces/entity.vllm.surface.prometheus_metrics https://schema.ai/vllm/metrics/entity.vllm.metric.num_requests_running https://schema.ai/vllm/metrics/entity.vllm.metric.num_requests_waiting https://schema.ai/vllm/metrics/entity.vllm.metric.num_requests_waiting_by_reason https://schema.ai/vllm/metrics/entity.vllm.metric.kv_cache_usage_perc https://schema.ai/vllm/metrics/entity.vllm.metric.num_preemptions https://schema.ai/vllm/metrics/entity.vllm.metric.engine_sleep_state https://schema.ai/vllm/metrics/entity.vllm.metric.corrupted_requests https://schema.ai/vllm/metrics/entity.vllm.metric.prefix_cache_queries https://schema.ai/vllm/metrics/entity.vllm.metric.prefix_cache_hits https://schema.ai/vllm/metrics/entity.vllm.metric.external_prefix_cache_queries https://schema.ai/vllm/metrics/entity.vllm.metric.external_prefix_cache_hits https://schema.ai/vllm/metrics/entity.vllm.metric.mm_cache_queries https://schema.ai/vllm/metrics/entity.vllm.metric.mm_cache_hits https://schema.ai/vllm/metrics/entity.vllm.metric.prompt_tokens https://schema.ai/vllm/metrics/entity.vllm.metric.prompt_tokens_by_source https://schema.ai/vllm/metrics/entity.vllm.metric.prompt_tokens_cached https://schema.ai/vllm/metrics/entity.vllm.metric.generation_tokens https://schema.ai/vllm/metrics/entity.vllm.metric.request_success https://schema.ai/vllm/metrics/entity.vllm.metric.request_prompt_tokens https://schema.ai/vllm/metrics/entity.vllm.metric.request_generation_tokens https://schema.ai/vllm/metrics/entity.vllm.metric.iteration_tokens_total https://schema.ai/vllm/metrics/entity.vllm.metric.request_max_num_generation_tokens https://schema.ai/vllm/metrics/entity.vllm.metric.request_params_n https://schema.ai/vllm/metrics/entity.vllm.metric.request_params_max_tokens https://schema.ai/vllm/metrics/entity.vllm.metric.time_to_first_token_seconds https://schema.ai/vllm/metrics/entity.vllm.metric.inter_token_latency_seconds https://schema.ai/vllm/metrics/entity.vllm.metric.request_time_per_output_token_seconds https://schema.ai/vllm/metrics/entity.vllm.metric.e2e_request_latency_seconds https://schema.ai/vllm/metrics/entity.vllm.metric.request_queue_time_seconds https://schema.ai/vllm/metrics/entity.vllm.metric.request_inference_time_seconds https://schema.ai/vllm/metrics/entity.vllm.metric.request_prefill_time_seconds https://schema.ai/vllm/metrics/entity.vllm.metric.request_decode_time_seconds https://schema.ai/vllm/metrics/entity.vllm.metric.request_prefill_kv_computed_tokens https://schema.ai/vllm/metrics/entity.vllm.metric.kv_block_lifetime_seconds https://schema.ai/vllm/metrics/entity.vllm.metric.kv_block_idle_before_evict_seconds https://schema.ai/vllm/metrics/entity.vllm.metric.kv_block_reuse_gap_seconds https://schema.ai/vllm/metrics/entity.vllm.metric.lora_requests_info https://schema.ai/vllm/metrics/entity.vllm.metric.cache_config_info https://schema.ai/vllm/metrics/entity.vllm.metric.spec_decode_num_drafts https://schema.ai/vllm/metrics/entity.vllm.metric.spec_decode_num_draft_tokens https://schema.ai/vllm/metrics/entity.vllm.metric.spec_decode_num_accepted_tokens https://schema.ai/vllm/metrics/entity.vllm.metric.spec_decode_num_accepted_tokens_per_pos https://schema.ai/vllm/metrics/entity.vllm.metric.estimated_flops_per_gpu_total https://schema.ai/vllm/metrics/entity.vllm.metric.estimated_read_bytes_per_gpu_total https://schema.ai/vllm/metrics/entity.vllm.metric.estimated_write_bytes_per_gpu_total https://schema.ai/vllm/settings/entity.vllm.parameter.is_encoder_decoder https://schema.ai/vllm/settings/entity.vllm.parameter.runner_type https://schema.ai/vllm/settings/entity.vllm.parameter.max_num_scheduled_tokens https://schema.ai/vllm/settings/entity.vllm.parameter.max_num_partial_prefills https://schema.ai/vllm/settings/entity.vllm.parameter.max_long_partial_prefills https://schema.ai/vllm/settings/entity.vllm.parameter.long_prefill_token_threshold https://schema.ai/vllm/settings/entity.vllm.parameter.enable_chunked_prefill https://schema.ai/vllm/settings/entity.vllm.parameter.is_multimodal_model https://schema.ai/vllm/settings/entity.vllm.parameter.max_num_encoder_input_tokens https://schema.ai/vllm/settings/entity.vllm.parameter.encoder_cache_size https://schema.ai/vllm/settings/entity.vllm.parameter.policy https://schema.ai/vllm/settings/entity.vllm.parameter.disable_chunked_mm_input https://schema.ai/vllm/settings/entity.vllm.parameter.scheduler_cls https://schema.ai/vllm/settings/entity.vllm.parameter.disable_hybrid_kv_cache_manager https://schema.ai/vllm/settings/entity.vllm.parameter.scheduler_reserve_full_isl https://schema.ai/vllm/settings/entity.vllm.parameter.async_scheduling https://schema.ai/vllm/settings/entity.vllm.parameter.stream_interval https://schema.ai/vllm/claims/claim.vllm.metric.request_decode_time_seconds.v0230.decode_interval_semantics https://schema.ai/vllm/claims/claim.vllm.metric.prefix_cache_queries.v0230.prometheus_counter https://schema.ai/vllm/claims/claim.vllm.metric.prefix_cache_hits.v0230.prometheus_counter https://schema.ai/vllm/claims/claim.vllm.metric.prefix_cache_queries.v0230.observes_prefix_cache_reuse https://schema.ai/vllm/claims/claim.vllm.metric.prefix_cache_hits.v0230.observes_prefix_cache_reuse https://schema.ai/vllm/claims/claim.vllm.metric.prefix_cache_hits.v0230.hit_rate_interpretation https://schema.ai/vllm/claims/claim.vllm.metric.num_requests_running.v0230.prometheus_gauge https://schema.ai/vllm/claims/claim.vllm.metric.num_requests_running.v0230.observes_scheduler_queue_state https://schema.ai/vllm/claims/claim.vllm.metric.num_requests_waiting.v0230.prometheus_gauge https://schema.ai/vllm/claims/claim.vllm.metric.num_requests_waiting_by_reason.v0230.reason_labels https://schema.ai/vllm/claims/claim.vllm.metric.kv_cache_usage_perc.v0230.prometheus_gauge https://schema.ai/vllm/claims/claim.vllm.metric.kv_cache_usage_perc.v0230.observes_kv_pressure https://schema.ai/vllm/claims/claim.vllm.metric.num_preemptions.v0230.prometheus_counter https://schema.ai/vllm/claims/claim.vllm.metric.num_preemptions.v0230.observes_preemption_pressure https://schema.ai/vllm/sources/source.vllm.code.config_scheduler.v0230 https://schema.ai/vllm/sources/source.vllm.code.kv_connector_v1_metrics.v0230 https://schema.ai/vllm/sources/source.vllm.code.v1_metrics_loggers.v0230 https://schema.ai/vllm/sources/source.vllm.code.v1_metrics_perf.v0230 https://schema.ai/vllm/sources/source.vllm.code.v1_metrics_stats.v0230 https://schema.ai/vllm/sources/source.vllm.code.v1_spec_decode_metrics.v0230 https://schema.ai/vllm/sources/source.vllm.docs.design_metrics.v0230 https://schema.ai/vllm/topics/topic.vllm.scheduler-kv-pressure-metrics https://schema.ai/vllm/topics/topic.vllm.prefix-cache-effectiveness https://schema.ai/vllm/scenarios/scenario.vllm.exposition-total-name https://schema.ai/vllm/scenarios/scenario.vllm.prefix-cache-metric-of-interest