{
  "rows": [
    {
      "case_id": "utf8-write",
      "split": "development",
      "class": "answerable",
      "retrieved_chunk_ids": [
        "path-read-3-13:3f86296ba6f04cc2",
        "path-read:1f1129b8d1011a82"
      ],
      "annotated_target_chunk_ids": [
        "path-write:b71a2d18a53d86bf"
      ],
      "retrieval": {
        "hit": false,
        "recall": 0.0,
        "reciprocal_rank": 0.0
      },
      "reply_status": "not_in_docs",
      "expected_status": "answered",
      "status_matches_annotation": false,
      "target_source_alignment": false,
      "error": null,
      "automatic_contracts_are_not_faithfulness": true,
      "human_semantic_verdict": "unreviewed",
      "human_review_required": true,
      "usage_metadata": {
        "input_tokens": 516,
        "output_tokens": 44,
        "total_tokens": 560
      },
      "response_metadata": {
        "model": "gemma4:e2b",
        "created_at": "2026-10-23T00:20:49.666594Z",
        "done": true,
        "done_reason": "stop",
        "total_duration": 9543054541,
        "load_duration": 5137902583,
        "prompt_eval_count": 516,
        "prompt_eval_duration": 274067000,
        "eval_count": 44,
        "eval_duration": 550648000,
        "logprobs": null,
        "model_name": "gemma4:e2b",
        "model_provider": "ollama"
      },
      "latency_seconds": 9.652761,
      "latency_scope": "one whole query including fresh query embedding when the fresh adapter is used; one-off document embedding/index build separate"
    },
    {
      "case_id": "replace-file",
      "split": "development",
      "class": "answerable",
      "retrieved_chunk_ids": [
        "path-write:b71a2d18a53d86bf",
        "path-read-3-13:3f86296ba6f04cc2"
      ],
      "annotated_target_chunk_ids": [
        "path-write:b71a2d18a53d86bf"
      ],
      "retrieval": {
        "hit": true,
        "recall": 1.0,
        "reciprocal_rank": 1.0
      },
      "reply_status": "answered",
      "expected_status": "answered",
      "status_matches_annotation": true,
      "target_source_alignment": true,
      "error": null,
      "automatic_contracts_are_not_faithfulness": true,
      "human_semantic_verdict": "unreviewed",
      "human_review_required": true,
      "usage_metadata": {
        "input_tokens": 497,
        "output_tokens": 39,
        "total_tokens": 536
      },
      "response_metadata": {
        "model": "gemma4:e2b",
        "created_at": "2026-10-23T00:20:54.058697Z",
        "done": true,
        "done_reason": "stop",
        "total_duration": 4290718334,
        "load_duration": 239142584,
        "prompt_eval_count": 497,
        "prompt_eval_duration": 256473000,
        "eval_count": 39,
        "eval_duration": 563510000,
        "logprobs": null,
        "model_name": "gemma4:e2b",
        "model_provider": "ollama"
      },
      "latency_seconds": 4.3891,
      "latency_scope": "one whole query including fresh query embedding when the fresh adapter is used; one-off document embedding/index build separate"
    },
    {
      "case_id": "json-result-type",
      "split": "development",
      "class": "answerable",
      "retrieved_chunk_ids": [
        "json-text:8a17e251a41ccb3b",
        "path-read-3-13:3f86296ba6f04cc2"
      ],
      "annotated_target_chunk_ids": [
        "json-text:8a17e251a41ccb3b"
      ],
      "retrieval": {
        "hit": true,
        "recall": 1.0,
        "reciprocal_rank": 1.0
      },
      "reply_status": "answered",
      "expected_status": "answered",
      "status_matches_annotation": true,
      "target_source_alignment": true,
      "error": null,
      "automatic_contracts_are_not_faithfulness": true,
      "human_semantic_verdict": "unreviewed",
      "human_review_required": true,
      "usage_metadata": {
        "input_tokens": 475,
        "output_tokens": 38,
        "total_tokens": 513
      },
      "response_metadata": {
        "model": "gemma4:e2b",
        "created_at": "2026-10-23T00:20:58.072658Z",
        "done": true,
        "done_reason": "stop",
        "total_duration": 3911592167,
        "load_duration": 233797250,
        "prompt_eval_count": 475,
        "prompt_eval_duration": 230309000,
        "eval_count": 38,
        "eval_duration": 459409000,
        "logprobs": null,
        "model_name": "gemma4:e2b",
        "model_provider": "ollama"
      },
      "latency_seconds": 4.017886,
      "latency_scope": "one whole query including fresh query embedding when the fresh adapter is used; one-off document embedding/index build separate"
    },
    {
      "case_id": "backup-schedule",
      "split": "development",
      "class": "unsupported",
      "retrieved_chunk_ids": [
        "path-read:1f1129b8d1011a82",
        "path-read-3-13:3f86296ba6f04cc2"
      ],
      "annotated_target_chunk_ids": [],
      "retrieval": null,
      "reply_status": "not_in_docs",
      "expected_status": "not_in_docs",
      "status_matches_annotation": true,
      "target_source_alignment": null,
      "error": null,
      "automatic_contracts_are_not_faithfulness": true,
      "human_semantic_verdict": "unreviewed",
      "human_review_required": true,
      "usage_metadata": {
        "input_tokens": 491,
        "output_tokens": 40,
        "total_tokens": 531
      },
      "response_metadata": {
        "model": "gemma4:e2b",
        "created_at": "2026-10-23T00:21:02.558949Z",
        "done": true,
        "done_reason": "stop",
        "total_duration": 4378516958,
        "load_duration": 232006750,
        "prompt_eval_count": 491,
        "prompt_eval_duration": 262749000,
        "eval_count": 40,
        "eval_duration": 544441000,
        "logprobs": null,
        "model_name": "gemma4:e2b",
        "model_provider": "ollama"
      },
      "latency_seconds": 4.482251,
      "latency_scope": "one whole query including fresh query embedding when the fresh adapter is used; one-off document embedding/index build separate"
    },
    {
      "case_id": "read-newline-312",
      "split": "held_out",
      "class": "version_specific",
      "retrieved_chunk_ids": [
        "path-read-3-13:3f86296ba6f04cc2",
        "path-read:1f1129b8d1011a82"
      ],
      "annotated_target_chunk_ids": [
        "path-read:1f1129b8d1011a82"
      ],
      "retrieval": {
        "hit": true,
        "recall": 1.0,
        "reciprocal_rank": 0.5
      },
      "reply_status": "not_in_docs",
      "expected_status": "answered",
      "status_matches_annotation": false,
      "target_source_alignment": false,
      "error": null,
      "automatic_contracts_are_not_faithfulness": true,
      "human_semantic_verdict": "unreviewed",
      "human_review_required": true,
      "usage_metadata": {
        "input_tokens": 554,
        "output_tokens": 59,
        "total_tokens": 613
      },
      "response_metadata": {
        "model": "gemma4:e2b",
        "created_at": "2026-10-23T00:21:07.362949Z",
        "done": true,
        "done_reason": "stop",
        "total_duration": 4699982625,
        "load_duration": 232594166,
        "prompt_eval_count": 554,
        "prompt_eval_duration": 63858000,
        "eval_count": 59,
        "eval_duration": 734530000,
        "logprobs": null,
        "model_name": "gemma4:e2b",
        "model_provider": "ollama"
      },
      "latency_seconds": 4.811534,
      "latency_scope": "one whole query including fresh query embedding when the fresh adapter is used; one-off document embedding/index build separate"
    },
    {
      "case_id": "read-newline-313",
      "split": "held_out",
      "class": "version_specific",
      "retrieved_chunk_ids": [
        "path-read-3-13:3f86296ba6f04cc2",
        "path-read:1f1129b8d1011a82"
      ],
      "annotated_target_chunk_ids": [
        "path-read-3-13:3f86296ba6f04cc2"
      ],
      "retrieval": {
        "hit": true,
        "recall": 1.0,
        "reciprocal_rank": 1.0
      },
      "reply_status": "answered",
      "expected_status": "answered",
      "status_matches_annotation": true,
      "target_source_alignment": true,
      "error": null,
      "automatic_contracts_are_not_faithfulness": true,
      "human_semantic_verdict": "unreviewed",
      "human_review_required": true,
      "usage_metadata": {
        "input_tokens": 553,
        "output_tokens": 60,
        "total_tokens": 613
      },
      "response_metadata": {
        "model": "gemma4:e2b",
        "created_at": "2026-10-23T00:21:12.180981Z",
        "done": true,
        "done_reason": "stop",
        "total_duration": 4704866083,
        "load_duration": 229715583,
        "prompt_eval_count": 553,
        "prompt_eval_duration": 61671000,
        "eval_count": 60,
        "eval_duration": 693529000,
        "logprobs": null,
        "model_name": "gemma4:e2b",
        "model_provider": "ollama"
      },
      "latency_seconds": 4.817051,
      "latency_scope": "one whole query including fresh query embedding when the fresh adapter is used; one-off document embedding/index build separate"
    },
    {
      "case_id": "json-throughput",
      "split": "held_out",
      "class": "unsupported",
      "retrieved_chunk_ids": [
        "json-text:8a17e251a41ccb3b",
        "path-read-3-13:3f86296ba6f04cc2"
      ],
      "annotated_target_chunk_ids": [],
      "retrieval": null,
      "reply_status": "not_in_docs",
      "expected_status": "not_in_docs",
      "status_matches_annotation": true,
      "target_source_alignment": null,
      "error": null,
      "automatic_contracts_are_not_faithfulness": true,
      "human_semantic_verdict": "unreviewed",
      "human_review_required": true,
      "usage_metadata": {
        "input_tokens": 500,
        "output_tokens": 41,
        "total_tokens": 541
      },
      "response_metadata": {
        "model": "gemma4:e2b",
        "created_at": "2026-10-23T00:21:16.731681Z",
        "done": true,
        "done_reason": "stop",
        "total_duration": 4443664000,
        "load_duration": 229657958,
        "prompt_eval_count": 500,
        "prompt_eval_duration": 257488999,
        "eval_count": 41,
        "eval_duration": 518532000,
        "logprobs": null,
        "model_name": "gemma4:e2b",
        "model_provider": "ollama"
      },
      "latency_seconds": 4.551986,
      "latency_scope": "one whole query including fresh query embedding when the fresh adapter is used; one-off document embedding/index build separate"
    },
    {
      "case_id": "create-parents",
      "split": "held_out",
      "class": "unsupported",
      "retrieved_chunk_ids": [
        "path-read-3-13:3f86296ba6f04cc2",
        "path-write:b71a2d18a53d86bf"
      ],
      "annotated_target_chunk_ids": [],
      "retrieval": null,
      "reply_status": null,
      "expected_status": "not_in_docs",
      "status_matches_annotation": null,
      "target_source_alignment": null,
      "error": {
        "types": [
          "ValueError",
          "Exception"
        ],
        "message": "Structured reply could not be parsed; raw reply retained, no repair"
      },
      "automatic_contracts_are_not_faithfulness": true,
      "human_semantic_verdict": "unreviewed",
      "human_review_required": true,
      "usage_metadata": {
        "input_tokens": 227,
        "output_tokens": 400,
        "total_tokens": 627
      },
      "response_metadata": {
        "model": "gemma4:e2b",
        "created_at": "2026-10-23T00:21:21.724875Z",
        "done": true,
        "done_reason": "length",
        "total_duration": 4881118750,
        "load_duration": 228535416,
        "prompt_eval_count": 227,
        "prompt_eval_duration": 194719000,
        "eval_count": 400,
        "eval_duration": 4415164000,
        "logprobs": null,
        "model_name": "gemma4:e2b",
        "model_provider": "ollama"
      },
      "latency_seconds": 4.987931,
      "latency_scope": "one whole query including fresh query embedding when the fresh adapter is used; one-off document embedding/index build separate"
    }
  ],
  "by_split": {
    "development": {
      "cases": 4,
      "supported": 3,
      "unsupported": 1,
      "retrieval_measurements": 3,
      "retrieval_unmeasured_errors": 0,
      "target_hits_at_2": 2,
      "mean_target_reciprocal_rank_at_2": 0.6666666666666666,
      "status_matches": 3,
      "false_answer_on_unsupported": 0,
      "false_abstention_on_supported": 1,
      "missing_or_error_outputs": 0,
      "human_semantics_reviewed": 0,
      "scope": "One trial per public teaching case, not a general accuracy estimate."
    },
    "held_out": {
      "cases": 4,
      "supported": 2,
      "unsupported": 2,
      "retrieval_measurements": 2,
      "retrieval_unmeasured_errors": 0,
      "target_hits_at_2": 2,
      "mean_target_reciprocal_rank_at_2": 0.75,
      "status_matches": 2,
      "false_answer_on_unsupported": 0,
      "false_abstention_on_supported": 1,
      "missing_or_error_outputs": 1,
      "human_semantics_reviewed": 0,
      "scope": "One trial per public teaching case, not a general accuracy estimate."
    }
  },
  "retrieval_null_rule": "Unsupported or unmeasured error→null, never perfect.",
  "metric_note": "One annotated target per supported case:recall@2equalsHit@2; these are not independent measurements."
}
