{"before": {"schema_version": 2, "lab_version": "0.2.0", "suite": "extraction", "split": "holdout", "created_at": "2026-10-01T20:51:14.682112+00:00", "candidate": {"name": "baseline", "kind": "local_rules", "model": null, "source_hash": "7665f5dde4f1c1a9ea008dbca97d9c101a4af0510720b6f00d5da49ad6c12e20"}, "dataset_hash": "6d0b67864847a451dbbac3bf6491035a711177f22883b6f77d54c2722cb61c03", "grader_hash": "7f4a79e5cafae9bdeef6d464191795315b45e142b4ddfb450f758a0b919e8d4c", "settings": {"allow_extra_fields": false}, "threshold": 0.8, "score": 0.3, "passed": 3, "failed": 7, "total": 10, "passed_gate": false, "gate_policy": {"critical_tags": [], "min_slices": {}, "fail_on_regression": false}, "gates": [{"name": "overall", "passed": false, "detail": "Overall score >= 80%"}], "slices": {"case": {"passed": 0, "total": 1, "accuracy": 0.0}, "missing": {"passed": 1, "total": 2, "accuracy": 0.5}, "negation": {"passed": 0, "total": 1, "accuracy": 0.0}, "plural": {"passed": 0, "total": 1, "accuracy": 0.0}, "priority": {"passed": 0, "total": 4, "accuracy": 0.0}, "typical": {"passed": 2, "total": 2, "accuracy": 1.0}, "vocabulary": {"passed": 0, "total": 1, "accuracy": 0.0}, "word-number": {"passed": 0, "total": 3, "accuracy": 0.0}}, "results": [{"id": "extract-holdout-01", "input": "Order S-922: 6 monitors.", "expected": {"order_id": "S-922", "quantity": 6, "item": "monitor", "priority": "normal"}, "tags": ["typical"], "case_id": "extract-holdout-01", "trial": 1, "actual": {"order_id": "S-922", "quantity": 6, "item": "monitor", "priority": "normal"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"S-922\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 6"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"monitor\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.007, "grader_latency_ms": 0.011, "latency_ms": 0.031}, {"id": "extract-holdout-02", "input": "Order T-023: four webcams, no rush.", "expected": {"order_id": "T-023", "quantity": 4, "item": "webcam", "priority": "low"}, "tags": ["word-number", "priority"], "case_id": "extract-holdout-02", "trial": 1, "actual": {"order_id": "T-023", "quantity": null, "item": null, "priority": "normal"}, "passed": false, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"T-023\""}, {"name": "field:quantity", "passed": false, "detail": "Expected quantity = 4"}, {"name": "field:item", "passed": false, "detail": "Expected item = \"webcam\""}, {"name": "field:priority", "passed": false, "detail": "Expected priority = \"low\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.003, "grader_latency_ms": 0.008, "latency_ms": 0.021}, {"id": "extract-holdout-03", "input": "Order U-124: six mice.", "expected": {"order_id": "U-124", "quantity": 6, "item": "mouse", "priority": "normal"}, "tags": ["word-number", "plural"], "case_id": "extract-holdout-03", "trial": 1, "actual": {"order_id": "U-124", "quantity": null, "item": null, "priority": "normal"}, "passed": false, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"U-124\""}, {"name": "field:quantity", "passed": false, "detail": "Expected quantity = 6"}, {"name": "field:item", "passed": false, "detail": "Expected item = \"mouse\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.002, "grader_latency_ms": 0.008, "latency_ms": 0.019}, {"id": "extract-holdout-04", "input": "Order v-225: 8 CABLES, urgent.", "expected": {"order_id": "V-225", "quantity": 8, "item": "cable", "priority": "high"}, "tags": ["case", "priority"], "case_id": "extract-holdout-04", "trial": 1, "actual": {"order_id": null, "quantity": null, "item": null, "priority": "high"}, "passed": false, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": false, "detail": "Expected order_id = \"V-225\""}, {"name": "field:quantity", "passed": false, "detail": "Expected quantity = 8"}, {"name": "field:item", "passed": false, "detail": "Expected item = \"cable\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"high\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.003, "grader_latency_ms": 0.007, "latency_ms": 0.018}, {"id": "extract-holdout-05", "input": "Order W-326: 2 printers.", "expected": {"order_id": "W-326", "quantity": 2, "item": "printer", "priority": "normal"}, "tags": ["vocabulary"], "case_id": "extract-holdout-05", "trial": 1, "actual": {"order_id": "W-326", "quantity": null, "item": null, "priority": "normal"}, "passed": false, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"W-326\""}, {"name": "field:quantity", "passed": false, "detail": "Expected quantity = 2"}, {"name": "field:item", "passed": false, "detail": "Expected item = \"printer\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.002, "grader_latency_ms": 0.007, "latency_ms": 0.017}, {"id": "extract-holdout-06", "input": "Order X-427: 3 keyboards, low priority.", "expected": {"order_id": "X-427", "quantity": 3, "item": "keyboard", "priority": "low"}, "tags": ["priority"], "case_id": "extract-holdout-06", "trial": 1, "actual": {"order_id": "X-427", "quantity": 3, "item": "keyboard", "priority": "normal"}, "passed": false, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"X-427\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 3"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"keyboard\""}, {"name": "field:priority", "passed": false, "detail": "Expected priority = \"low\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.003, "grader_latency_ms": 0.007, "latency_ms": 0.018}, {"id": "extract-holdout-07", "input": "Order Y-528: 1 headset, not urgent.", "expected": {"order_id": "Y-528", "quantity": 1, "item": "headset", "priority": "low"}, "tags": ["negation", "priority"], "case_id": "extract-holdout-07", "trial": 1, "actual": {"order_id": "Y-528", "quantity": 1, "item": "headset", "priority": "high"}, "passed": false, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"Y-528\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 1"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"headset\""}, {"name": "field:priority", "passed": false, "detail": "Expected priority = \"low\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.003, "grader_latency_ms": 0.044, "latency_ms": 0.056}, {"id": "extract-holdout-08", "input": "Please send five monitors ASAP.", "expected": {"order_id": null, "quantity": 5, "item": "monitor", "priority": "high"}, "tags": ["missing", "word-number"], "case_id": "extract-holdout-08", "trial": 1, "actual": {"order_id": null, "quantity": null, "item": null, "priority": "normal"}, "passed": false, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = null"}, {"name": "field:quantity", "passed": false, "detail": "Expected quantity = 5"}, {"name": "field:item", "passed": false, "detail": "Expected item = \"monitor\""}, {"name": "field:priority", "passed": false, "detail": "Expected priority = \"high\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.002, "grader_latency_ms": 0.008, "latency_ms": 0.019}, {"id": "extract-holdout-09", "input": "Order Z-629 was recorded.", "expected": {"order_id": "Z-629", "quantity": null, "item": null, "priority": "normal"}, "tags": ["missing"], "case_id": "extract-holdout-09", "trial": 1, "actual": {"order_id": "Z-629", "quantity": null, "item": null, "priority": "normal"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"Z-629\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = null"}, {"name": "field:item", "passed": true, "detail": "Expected item = null"}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.002, "grader_latency_ms": 0.008, "latency_ms": 0.018}, {"id": "extract-holdout-10", "input": "Order A-730: 7 webcams, normal priority.", "expected": {"order_id": "A-730", "quantity": 7, "item": "webcam", "priority": "normal"}, "tags": ["typical"], "case_id": "extract-holdout-10", "trial": 1, "actual": {"order_id": "A-730", "quantity": 7, "item": "webcam", "priority": "normal"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"A-730\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 7"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"webcam\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.003, "grader_latency_ms": 0.007, "latency_ms": 0.018}], "trials": 1, "case_count": 10, "pair_order": "original", "execution_errors": 0, "data_notice": "Bundled cases are synthetic teaching data under MIT-0. Custom cases may have different provenance.", "metrics": {"checks": {"json_object": {"passed": 10, "total": 10, "rate": 1.0}, "schema": {"passed": 10, "total": 10, "rate": 1.0}, "field:order_id": {"passed": 9, "total": 10, "rate": 0.9}, "field:quantity": {"passed": 5, "total": 10, "rate": 0.5}, "field:item": {"passed": 5, "total": 10, "rate": 0.5}, "field:priority": {"passed": 6, "total": 10, "rate": 0.6}}, "measurements": {}, "candidate_latency_ms": {"median": 0.003, "p95": 0.007}, "usage": {}, "usage_coverage": 0, "trial_scores": [0.3], "unstable_cases": 0}}, "after": {"schema_version": 2, "lab_version": "0.2.0", "suite": "extraction", "split": "holdout", "created_at": "2026-10-01T20:51:14.683451+00:00", "candidate": {"name": "improved", "kind": "local_rules", "model": null, "source_hash": "7665f5dde4f1c1a9ea008dbca97d9c101a4af0510720b6f00d5da49ad6c12e20"}, "dataset_hash": "6d0b67864847a451dbbac3bf6491035a711177f22883b6f77d54c2722cb61c03", "grader_hash": "7f4a79e5cafae9bdeef6d464191795315b45e142b4ddfb450f758a0b919e8d4c", "settings": {"allow_extra_fields": false}, "threshold": 0.8, "score": 0.8, "passed": 8, "failed": 2, "total": 10, "passed_gate": true, "gate_policy": {"critical_tags": [], "min_slices": {}, "fail_on_regression": false}, "gates": [{"name": "overall", "passed": true, "detail": "Overall score >= 80%"}], "slices": {"case": {"passed": 1, "total": 1, "accuracy": 1.0}, "missing": {"passed": 2, "total": 2, "accuracy": 1.0}, "negation": {"passed": 1, "total": 1, "accuracy": 1.0}, "plural": {"passed": 0, "total": 1, "accuracy": 0.0}, "priority": {"passed": 4, "total": 4, "accuracy": 1.0}, "typical": {"passed": 2, "total": 2, "accuracy": 1.0}, "vocabulary": {"passed": 0, "total": 1, "accuracy": 0.0}, "word-number": {"passed": 2, "total": 3, "accuracy": 0.6666666666666666}}, "results": [{"id": "extract-holdout-01", "input": "Order S-922: 6 monitors.", "expected": {"order_id": "S-922", "quantity": 6, "item": "monitor", "priority": "normal"}, "tags": ["typical"], "case_id": "extract-holdout-01", "trial": 1, "actual": {"order_id": "S-922", "quantity": 6, "item": "monitor", "priority": "normal"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"S-922\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 6"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"monitor\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.012, "grader_latency_ms": 0.012, "latency_ms": 0.04}, {"id": "extract-holdout-02", "input": "Order T-023: four webcams, no rush.", "expected": {"order_id": "T-023", "quantity": 4, "item": "webcam", "priority": "low"}, "tags": ["word-number", "priority"], "case_id": "extract-holdout-02", "trial": 1, "actual": {"order_id": "T-023", "quantity": 4, "item": "webcam", "priority": "low"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"T-023\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 4"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"webcam\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"low\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.006, "grader_latency_ms": 0.009, "latency_ms": 0.025}, {"id": "extract-holdout-03", "input": "Order U-124: six mice.", "expected": {"order_id": "U-124", "quantity": 6, "item": "mouse", "priority": "normal"}, "tags": ["word-number", "plural"], "case_id": "extract-holdout-03", "trial": 1, "actual": {"order_id": "U-124", "quantity": null, "item": null, "priority": "normal"}, "passed": false, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"U-124\""}, {"name": "field:quantity", "passed": false, "detail": "Expected quantity = 6"}, {"name": "field:item", "passed": false, "detail": "Expected item = \"mouse\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.005, "grader_latency_ms": 0.008, "latency_ms": 0.022}, {"id": "extract-holdout-04", "input": "Order v-225: 8 CABLES, urgent.", "expected": {"order_id": "V-225", "quantity": 8, "item": "cable", "priority": "high"}, "tags": ["case", "priority"], "case_id": "extract-holdout-04", "trial": 1, "actual": {"order_id": "V-225", "quantity": 8, "item": "cable", "priority": "high"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"V-225\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 8"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"cable\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"high\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.005, "grader_latency_ms": 0.008, "latency_ms": 0.021}, {"id": "extract-holdout-05", "input": "Order W-326: 2 printers.", "expected": {"order_id": "W-326", "quantity": 2, "item": "printer", "priority": "normal"}, "tags": ["vocabulary"], "case_id": "extract-holdout-05", "trial": 1, "actual": {"order_id": "W-326", "quantity": null, "item": null, "priority": "normal"}, "passed": false, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"W-326\""}, {"name": "field:quantity", "passed": false, "detail": "Expected quantity = 2"}, {"name": "field:item", "passed": false, "detail": "Expected item = \"printer\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.004, "grader_latency_ms": 0.007, "latency_ms": 0.02}, {"id": "extract-holdout-06", "input": "Order X-427: 3 keyboards, low priority.", "expected": {"order_id": "X-427", "quantity": 3, "item": "keyboard", "priority": "low"}, "tags": ["priority"], "case_id": "extract-holdout-06", "trial": 1, "actual": {"order_id": "X-427", "quantity": 3, "item": "keyboard", "priority": "low"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"X-427\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 3"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"keyboard\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"low\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.004, "grader_latency_ms": 0.007, "latency_ms": 0.019}, {"id": "extract-holdout-07", "input": "Order Y-528: 1 headset, not urgent.", "expected": {"order_id": "Y-528", "quantity": 1, "item": "headset", "priority": "low"}, "tags": ["negation", "priority"], "case_id": "extract-holdout-07", "trial": 1, "actual": {"order_id": "Y-528", "quantity": 1, "item": "headset", "priority": "low"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"Y-528\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 1"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"headset\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"low\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.004, "grader_latency_ms": 0.007, "latency_ms": 0.019}, {"id": "extract-holdout-08", "input": "Please send five monitors ASAP.", "expected": {"order_id": null, "quantity": 5, "item": "monitor", "priority": "high"}, "tags": ["missing", "word-number"], "case_id": "extract-holdout-08", "trial": 1, "actual": {"order_id": null, "quantity": 5, "item": "monitor", "priority": "high"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = null"}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 5"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"monitor\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"high\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.005, "grader_latency_ms": 0.008, "latency_ms": 0.022}, {"id": "extract-holdout-09", "input": "Order Z-629 was recorded.", "expected": {"order_id": "Z-629", "quantity": null, "item": null, "priority": "normal"}, "tags": ["missing"], "case_id": "extract-holdout-09", "trial": 1, "actual": {"order_id": "Z-629", "quantity": null, "item": null, "priority": "normal"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"Z-629\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = null"}, {"name": "field:item", "passed": true, "detail": "Expected item = null"}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.004, "grader_latency_ms": 0.008, "latency_ms": 0.02}, {"id": "extract-holdout-10", "input": "Order A-730: 7 webcams, normal priority.", "expected": {"order_id": "A-730", "quantity": 7, "item": "webcam", "priority": "normal"}, "tags": ["typical"], "case_id": "extract-holdout-10", "trial": 1, "actual": {"order_id": "A-730", "quantity": 7, "item": "webcam", "priority": "normal"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"A-730\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 7"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"webcam\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.005, "grader_latency_ms": 0.007, "latency_ms": 0.02}], "trials": 1, "case_count": 10, "pair_order": "original", "execution_errors": 0, "data_notice": "Bundled cases are synthetic teaching data under MIT-0. Custom cases may have different provenance.", "metrics": {"checks": {"json_object": {"passed": 10, "total": 10, "rate": 1.0}, "schema": {"passed": 10, "total": 10, "rate": 1.0}, "field:order_id": {"passed": 10, "total": 10, "rate": 1.0}, "field:quantity": {"passed": 8, "total": 10, "rate": 0.8}, "field:item": {"passed": 8, "total": 10, "rate": 0.8}, "field:priority": {"passed": 10, "total": 10, "rate": 1.0}}, "measurements": {}, "candidate_latency_ms": {"median": 0.005, "p95": 0.012}, "usage": {}, "usage_coverage": 0, "trial_scores": [0.8], "unstable_cases": 0}}, "delta": 0.5, "improved": 5, "regressed": 0, "changes": [{"id": "extract-holdout-01", "change": "unchanged", "before": {"id": "extract-holdout-01", "input": "Order S-922: 6 monitors.", "expected": {"order_id": "S-922", "quantity": 6, "item": "monitor", "priority": "normal"}, "tags": ["typical"], "case_id": "extract-holdout-01", "trial": 1, "actual": {"order_id": "S-922", "quantity": 6, "item": "monitor", "priority": "normal"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"S-922\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 6"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"monitor\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.007, "grader_latency_ms": 0.011, "latency_ms": 0.031}, "after": {"id": "extract-holdout-01", "input": "Order S-922: 6 monitors.", "expected": {"order_id": "S-922", "quantity": 6, "item": "monitor", "priority": "normal"}, "tags": ["typical"], "case_id": "extract-holdout-01", "trial": 1, "actual": {"order_id": "S-922", "quantity": 6, "item": "monitor", "priority": "normal"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"S-922\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 6"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"monitor\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.012, "grader_latency_ms": 0.012, "latency_ms": 0.04}}, {"id": "extract-holdout-02", "change": "improved", "before": {"id": "extract-holdout-02", "input": "Order T-023: four webcams, no rush.", "expected": {"order_id": "T-023", "quantity": 4, "item": "webcam", "priority": "low"}, "tags": ["word-number", "priority"], "case_id": "extract-holdout-02", "trial": 1, "actual": {"order_id": "T-023", "quantity": null, "item": null, "priority": "normal"}, "passed": false, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"T-023\""}, {"name": "field:quantity", "passed": false, "detail": "Expected quantity = 4"}, {"name": "field:item", "passed": false, "detail": "Expected item = \"webcam\""}, {"name": "field:priority", "passed": false, "detail": "Expected priority = \"low\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.003, "grader_latency_ms": 0.008, "latency_ms": 0.021}, "after": {"id": "extract-holdout-02", "input": "Order T-023: four webcams, no rush.", "expected": {"order_id": "T-023", "quantity": 4, "item": "webcam", "priority": "low"}, "tags": ["word-number", "priority"], "case_id": "extract-holdout-02", "trial": 1, "actual": {"order_id": "T-023", "quantity": 4, "item": "webcam", "priority": "low"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"T-023\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 4"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"webcam\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"low\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.006, "grader_latency_ms": 0.009, "latency_ms": 0.025}}, {"id": "extract-holdout-03", "change": "unchanged", "before": {"id": "extract-holdout-03", "input": "Order U-124: six mice.", "expected": {"order_id": "U-124", "quantity": 6, "item": "mouse", "priority": "normal"}, "tags": ["word-number", "plural"], "case_id": "extract-holdout-03", "trial": 1, "actual": {"order_id": "U-124", "quantity": null, "item": null, "priority": "normal"}, "passed": false, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"U-124\""}, {"name": "field:quantity", "passed": false, "detail": "Expected quantity = 6"}, {"name": "field:item", "passed": false, "detail": "Expected item = \"mouse\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.002, "grader_latency_ms": 0.008, "latency_ms": 0.019}, "after": {"id": "extract-holdout-03", "input": "Order U-124: six mice.", "expected": {"order_id": "U-124", "quantity": 6, "item": "mouse", "priority": "normal"}, "tags": ["word-number", "plural"], "case_id": "extract-holdout-03", "trial": 1, "actual": {"order_id": "U-124", "quantity": null, "item": null, "priority": "normal"}, "passed": false, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"U-124\""}, {"name": "field:quantity", "passed": false, "detail": "Expected quantity = 6"}, {"name": "field:item", "passed": false, "detail": "Expected item = \"mouse\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.005, "grader_latency_ms": 0.008, "latency_ms": 0.022}}, {"id": "extract-holdout-04", "change": "improved", "before": {"id": "extract-holdout-04", "input": "Order v-225: 8 CABLES, urgent.", "expected": {"order_id": "V-225", "quantity": 8, "item": "cable", "priority": "high"}, "tags": ["case", "priority"], "case_id": "extract-holdout-04", "trial": 1, "actual": {"order_id": null, "quantity": null, "item": null, "priority": "high"}, "passed": false, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": false, "detail": "Expected order_id = \"V-225\""}, {"name": "field:quantity", "passed": false, "detail": "Expected quantity = 8"}, {"name": "field:item", "passed": false, "detail": "Expected item = \"cable\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"high\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.003, "grader_latency_ms": 0.007, "latency_ms": 0.018}, "after": {"id": "extract-holdout-04", "input": "Order v-225: 8 CABLES, urgent.", "expected": {"order_id": "V-225", "quantity": 8, "item": "cable", "priority": "high"}, "tags": ["case", "priority"], "case_id": "extract-holdout-04", "trial": 1, "actual": {"order_id": "V-225", "quantity": 8, "item": "cable", "priority": "high"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"V-225\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 8"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"cable\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"high\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.005, "grader_latency_ms": 0.008, "latency_ms": 0.021}}, {"id": "extract-holdout-05", "change": "unchanged", "before": {"id": "extract-holdout-05", "input": "Order W-326: 2 printers.", "expected": {"order_id": "W-326", "quantity": 2, "item": "printer", "priority": "normal"}, "tags": ["vocabulary"], "case_id": "extract-holdout-05", "trial": 1, "actual": {"order_id": "W-326", "quantity": null, "item": null, "priority": "normal"}, "passed": false, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"W-326\""}, {"name": "field:quantity", "passed": false, "detail": "Expected quantity = 2"}, {"name": "field:item", "passed": false, "detail": "Expected item = \"printer\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.002, "grader_latency_ms": 0.007, "latency_ms": 0.017}, "after": {"id": "extract-holdout-05", "input": "Order W-326: 2 printers.", "expected": {"order_id": "W-326", "quantity": 2, "item": "printer", "priority": "normal"}, "tags": ["vocabulary"], "case_id": "extract-holdout-05", "trial": 1, "actual": {"order_id": "W-326", "quantity": null, "item": null, "priority": "normal"}, "passed": false, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"W-326\""}, {"name": "field:quantity", "passed": false, "detail": "Expected quantity = 2"}, {"name": "field:item", "passed": false, "detail": "Expected item = \"printer\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.004, "grader_latency_ms": 0.007, "latency_ms": 0.02}}, {"id": "extract-holdout-06", "change": "improved", "before": {"id": "extract-holdout-06", "input": "Order X-427: 3 keyboards, low priority.", "expected": {"order_id": "X-427", "quantity": 3, "item": "keyboard", "priority": "low"}, "tags": ["priority"], "case_id": "extract-holdout-06", "trial": 1, "actual": {"order_id": "X-427", "quantity": 3, "item": "keyboard", "priority": "normal"}, "passed": false, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"X-427\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 3"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"keyboard\""}, {"name": "field:priority", "passed": false, "detail": "Expected priority = \"low\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.003, "grader_latency_ms": 0.007, "latency_ms": 0.018}, "after": {"id": "extract-holdout-06", "input": "Order X-427: 3 keyboards, low priority.", "expected": {"order_id": "X-427", "quantity": 3, "item": "keyboard", "priority": "low"}, "tags": ["priority"], "case_id": "extract-holdout-06", "trial": 1, "actual": {"order_id": "X-427", "quantity": 3, "item": "keyboard", "priority": "low"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"X-427\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 3"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"keyboard\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"low\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.004, "grader_latency_ms": 0.007, "latency_ms": 0.019}}, {"id": "extract-holdout-07", "change": "improved", "before": {"id": "extract-holdout-07", "input": "Order Y-528: 1 headset, not urgent.", "expected": {"order_id": "Y-528", "quantity": 1, "item": "headset", "priority": "low"}, "tags": ["negation", "priority"], "case_id": "extract-holdout-07", "trial": 1, "actual": {"order_id": "Y-528", "quantity": 1, "item": "headset", "priority": "high"}, "passed": false, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"Y-528\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 1"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"headset\""}, {"name": "field:priority", "passed": false, "detail": "Expected priority = \"low\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.003, "grader_latency_ms": 0.044, "latency_ms": 0.056}, "after": {"id": "extract-holdout-07", "input": "Order Y-528: 1 headset, not urgent.", "expected": {"order_id": "Y-528", "quantity": 1, "item": "headset", "priority": "low"}, "tags": ["negation", "priority"], "case_id": "extract-holdout-07", "trial": 1, "actual": {"order_id": "Y-528", "quantity": 1, "item": "headset", "priority": "low"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"Y-528\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 1"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"headset\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"low\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.004, "grader_latency_ms": 0.007, "latency_ms": 0.019}}, {"id": "extract-holdout-08", "change": "improved", "before": {"id": "extract-holdout-08", "input": "Please send five monitors ASAP.", "expected": {"order_id": null, "quantity": 5, "item": "monitor", "priority": "high"}, "tags": ["missing", "word-number"], "case_id": "extract-holdout-08", "trial": 1, "actual": {"order_id": null, "quantity": null, "item": null, "priority": "normal"}, "passed": false, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = null"}, {"name": "field:quantity", "passed": false, "detail": "Expected quantity = 5"}, {"name": "field:item", "passed": false, "detail": "Expected item = \"monitor\""}, {"name": "field:priority", "passed": false, "detail": "Expected priority = \"high\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.002, "grader_latency_ms": 0.008, "latency_ms": 0.019}, "after": {"id": "extract-holdout-08", "input": "Please send five monitors ASAP.", "expected": {"order_id": null, "quantity": 5, "item": "monitor", "priority": "high"}, "tags": ["missing", "word-number"], "case_id": "extract-holdout-08", "trial": 1, "actual": {"order_id": null, "quantity": 5, "item": "monitor", "priority": "high"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = null"}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 5"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"monitor\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"high\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.005, "grader_latency_ms": 0.008, "latency_ms": 0.022}}, {"id": "extract-holdout-09", "change": "unchanged", "before": {"id": "extract-holdout-09", "input": "Order Z-629 was recorded.", "expected": {"order_id": "Z-629", "quantity": null, "item": null, "priority": "normal"}, "tags": ["missing"], "case_id": "extract-holdout-09", "trial": 1, "actual": {"order_id": "Z-629", "quantity": null, "item": null, "priority": "normal"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"Z-629\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = null"}, {"name": "field:item", "passed": true, "detail": "Expected item = null"}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.002, "grader_latency_ms": 0.008, "latency_ms": 0.018}, "after": {"id": "extract-holdout-09", "input": "Order Z-629 was recorded.", "expected": {"order_id": "Z-629", "quantity": null, "item": null, "priority": "normal"}, "tags": ["missing"], "case_id": "extract-holdout-09", "trial": 1, "actual": {"order_id": "Z-629", "quantity": null, "item": null, "priority": "normal"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"Z-629\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = null"}, {"name": "field:item", "passed": true, "detail": "Expected item = null"}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.004, "grader_latency_ms": 0.008, "latency_ms": 0.02}}, {"id": "extract-holdout-10", "change": "unchanged", "before": {"id": "extract-holdout-10", "input": "Order A-730: 7 webcams, normal priority.", "expected": {"order_id": "A-730", "quantity": 7, "item": "webcam", "priority": "normal"}, "tags": ["typical"], "case_id": "extract-holdout-10", "trial": 1, "actual": {"order_id": "A-730", "quantity": 7, "item": "webcam", "priority": "normal"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"A-730\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 7"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"webcam\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.003, "grader_latency_ms": 0.007, "latency_ms": 0.018}, "after": {"id": "extract-holdout-10", "input": "Order A-730: 7 webcams, normal priority.", "expected": {"order_id": "A-730", "quantity": 7, "item": "webcam", "priority": "normal"}, "tags": ["typical"], "case_id": "extract-holdout-10", "trial": 1, "actual": {"order_id": "A-730", "quantity": 7, "item": "webcam", "priority": "normal"}, "passed": true, "checks": [{"name": "json_object", "passed": true, "detail": "Output must be a JSON object."}, {"name": "schema", "passed": true, "detail": "Check required fields and types; a boolean is not an integer."}, {"name": "field:order_id", "passed": true, "detail": "Expected order_id = \"A-730\""}, {"name": "field:quantity", "passed": true, "detail": "Expected quantity = 7"}, {"name": "field:item", "passed": true, "detail": "Expected item = \"webcam\""}, {"name": "field:priority", "passed": true, "detail": "Expected priority = \"normal\""}], "error": null, "usage": {}, "metadata": {}, "candidate_latency_ms": 0.005, "grader_latency_ms": 0.007, "latency_ms": 0.02}}], "gates": [{"name": "overall", "passed": true, "detail": "Overall score >= 80%"}], "passed_gate": true}
