{ "schema": "MAKNOW_AIE007_LAB_FIXTURES_v1", "cases": [ { "id": "baseline_pass", "model_revision": "atlas-7b@abc123", "vllm_version": "0.x-pinned", "measured_config_hash": "cfg-a", "current_config_hash": "cfg-a", "hardware_profile": "2xGPU-profile-a", "tensor_parallel_size": 1, "pipeline_parallel_size": 1, "max_model_len": 8192, "max_num_batched_tokens": 4096, "max_num_seqs": 8, "measured_input_max": 2048, "measured_output_max": 512, "measured_concurrency_max": 8, "target_input": 1800, "target_output": 400, "target_concurrency": 8, "warmup_state": "WARM", "p95_latency_ms": 1600, "p95_slo_ms": 1800, "throughput_tps": 140, "min_throughput_tps": 120, "queue_p95_ms": 120, "error_count": 0, "timeout_count": 0, "security_boundary_recorded": true }, { "id": "outside_concurrency", "model_revision": "atlas-7b@abc123", "vllm_version": "0.x-pinned", "measured_config_hash": "cfg-a", "current_config_hash": "cfg-a", "hardware_profile": "2xGPU-profile-a", "tensor_parallel_size": 1, "pipeline_parallel_size": 1, "max_model_len": 8192, "max_num_batched_tokens": 4096, "max_num_seqs": 8, "measured_input_max": 2048, "measured_output_max": 512, "measured_concurrency_max": 8, "target_input": 1800, "target_output": 400, "target_concurrency": 12, "warmup_state": "WARM", "p95_latency_ms": 1600, "p95_slo_ms": 1800, "throughput_tps": 140, "min_throughput_tps": 120, "queue_p95_ms": 120, "error_count": 0, "timeout_count": 0, "security_boundary_recorded": true }, { "id": "config_changed", "model_revision": "atlas-7b@abc123", "vllm_version": "0.x-pinned", "measured_config_hash": "cfg-a", "current_config_hash": "cfg-b", "hardware_profile": "2xGPU-profile-a", "tensor_parallel_size": 2, "pipeline_parallel_size": 1, "max_model_len": 16384, "max_num_batched_tokens": 8192, "max_num_seqs": 8, "measured_input_max": 2048, "measured_output_max": 512, "measured_concurrency_max": 8, "target_input": 1800, "target_output": 400, "target_concurrency": 8, "warmup_state": "WARM", "p95_latency_ms": 1600, "p95_slo_ms": 1800, "throughput_tps": 140, "min_throughput_tps": 120, "queue_p95_ms": 120, "error_count": 0, "timeout_count": 0, "security_boundary_recorded": true }, { "id": "provenance_incomplete", "model_revision": "atlas-7b@abc123", "vllm_version": "", "measured_config_hash": "cfg-a", "current_config_hash": "cfg-a", "hardware_profile": "2xGPU-profile-a", "tensor_parallel_size": 1, "pipeline_parallel_size": 1, "max_model_len": 8192, "max_num_batched_tokens": 4096, "max_num_seqs": 8, "measured_input_max": 2048, "measured_output_max": 512, "measured_concurrency_max": 8, "target_input": 1800, "target_output": 400, "target_concurrency": 8, "warmup_state": "WARM", "p95_latency_ms": 1600, "p95_slo_ms": 1800, "throughput_tps": 140, "min_throughput_tps": 120, "queue_p95_ms": 120, "error_count": 0, "timeout_count": 0, "security_boundary_recorded": true }, { "id": "warmup_unknown", "model_revision": "atlas-7b@abc123", "vllm_version": "0.x-pinned", "measured_config_hash": "cfg-a", "current_config_hash": "cfg-a", "hardware_profile": "2xGPU-profile-a", "tensor_parallel_size": 1, "pipeline_parallel_size": 1, "max_model_len": 8192, "max_num_batched_tokens": 4096, "max_num_seqs": 8, "measured_input_max": 2048, "measured_output_max": 512, "measured_concurrency_max": 8, "target_input": 1800, "target_output": 400, "target_concurrency": 8, "warmup_state": "UNKNOWN", "p95_latency_ms": 1600, "p95_slo_ms": 1800, "throughput_tps": 140, "min_throughput_tps": 120, "queue_p95_ms": 120, "error_count": 0, "timeout_count": 0, "security_boundary_recorded": true }, { "id": "slo_failed", "model_revision": "atlas-7b@abc123", "vllm_version": "0.x-pinned", "measured_config_hash": "cfg-a", "current_config_hash": "cfg-a", "hardware_profile": "2xGPU-profile-a", "tensor_parallel_size": 1, "pipeline_parallel_size": 1, "max_model_len": 8192, "max_num_batched_tokens": 4096, "max_num_seqs": 8, "measured_input_max": 2048, "measured_output_max": 512, "measured_concurrency_max": 8, "target_input": 1800, "target_output": 400, "target_concurrency": 8, "warmup_state": "WARM", "p95_latency_ms": 2400, "p95_slo_ms": 1800, "throughput_tps": 105, "min_throughput_tps": 120, "queue_p95_ms": 600, "error_count": 1, "timeout_count": 1, "security_boundary_recorded": true }, { "id": "security_boundary_missing", "model_revision": "atlas-7b@abc123", "vllm_version": "0.x-pinned", "measured_config_hash": "cfg-a", "current_config_hash": "cfg-a", "hardware_profile": "2xGPU-profile-a", "tensor_parallel_size": 1, "pipeline_parallel_size": 1, "max_model_len": 8192, "max_num_batched_tokens": 4096, "max_num_seqs": 8, "measured_input_max": 2048, "measured_output_max": 512, "measured_concurrency_max": 8, "target_input": 1800, "target_output": 400, "target_concurrency": 8, "warmup_state": "WARM", "p95_latency_ms": 1600, "p95_slo_ms": 1800, "throughput_tps": 140, "min_throughput_tps": 120, "queue_p95_ms": 120, "error_count": 0, "timeout_count": 0, "security_boundary_recorded": false }, { "id": "transfer_larger_context", "model_revision": "atlas-7b@abc123", "vllm_version": "0.x-pinned", "measured_config_hash": "cfg-a", "current_config_hash": "cfg-a", "hardware_profile": "2xGPU-profile-a", "tensor_parallel_size": 1, "pipeline_parallel_size": 1, "max_model_len": 16384, "max_num_batched_tokens": 4096, "max_num_seqs": 8, "measured_input_max": 2048, "measured_output_max": 512, "measured_concurrency_max": 8, "target_input": 4096, "target_output": 1024, "target_concurrency": 12, "warmup_state": "WARM", "p95_latency_ms": 1600, "p95_slo_ms": 1800, "throughput_tps": 140, "min_throughput_tps": 120, "queue_p95_ms": 120, "error_count": 0, "timeout_count": 0, "security_boundary_recorded": true } ] }