EVALUATION.md

# Evaluate new predictions and replay the paper

Install `requirements.txt` in a Python 3.12+ environment. The verified runtime
and exact package versions are in `protocol/verified-runtime.json`. All commands
below run on CPU, use local files, and make no model/API calls. Run from this
dataset folder, with outputs outside it so the frozen manifest stays valid.

## New model predictions

Use only `image` and `question` as model inputs. Read the exact prompt and schema
from `protocol/prompt.json`. Obtain GQA inputs using `RECONSTRUCTION.md` first.
Write one JSON object per input in `predictions.jsonl`:

```json
{"item_id":"item-0001","answerable":false,"answer":null,"reason":"The needed value is not visible."}
```

That is a format example, not a prescribed answer for other inputs. Retain each
stable item ID. For a terminal malformed model output use:

```json
{"item_id":"item-0001","status":"invalid_schema","terminal_invalid":true,"parsed":null}
```

One row is required for every selected input. Do not submit both examples for
the same ID. Missing/duplicate/unexpected IDs or incomplete state groups cause
the scorer to fail rather than silently shrink the denominator. Invalid final
outputs count as failures. A valid answerable decision with a null answer can
pass decision scoring but fails answer correctness. Unanswerable decisions
must have a null answer and all valid outputs need a nonempty reason.

```bash
python scripts/score_hf_dataset.py --predictions ../predictions.jsonl \
  --sources plotqa clevr gqa --output ../scores.json
```

For a single subset, supply only its predictions and e.g. `--sources plotqa`.
Output includes exact counts, per-state failures, per-view decision accuracy,
all-decisions-correct groups, supported-answer accuracy, class-balanced failure,
and complete joint answer/abstention success. Chart answers use the paper's
conservative exact-rational parser (no numeric tolerance). Scene scoring first
checks exact rationals, then Unicode NFKC, case and whitespace normalization;
it does not infer synonyms. A percentage marker follows the historical parser
and is not automatically divided by 100.

The GQA flag in the metadata defines the same 376-group location sensitivity
stratum as the paper. The supplied saved-response replay reports both strata.

## Reproduce reported statistics

```bash
python scripts/replay_hf_statistics.py --output ../replay
```

This checks all 72,000 scoring projections, recomputes all 18 model/domain cells
and the paper's 10,000-draw paired group bootstrap intervals, compares the full
analysis JSON to the frozen expected result, regenerates tables and chart
diagnostics, verifies the 5,000 chart view-to-proof identity joins, and runs the
new-prediction scorer against all six saved configurations. It also checks the
finite-study counts and enumeration/SMT agreement from saved per-case records.
It does not regenerate the finite solver-study witnesses, raw generations or
historical serving traces. GQA question-ID tokens replace upstream question
text consistently in numerical replay; reconstruction checks the actual text.

The supplementary shortcut and answer-error counts can also be checked directly:

```bash
python scripts/verify_argument_evidence.py --inputs inputs \
  --finite-results evidence/finite/original-results.jsonl.gz \
  --solver-results evidence/finite/solver-results.jsonl.gz \
  --output ../supplementary-counts.json
```

```bash
python scripts/verify_hf_dataset.py --workers 8 --output ../verification.json
```

This verifies the frozen file manifest, loads all three configurations through
Hugging Face Datasets, decodes and hashes all 9,000 bundled images, checks every
state group and unique source-image identity, verifies all 1,000 complete chart
proofs against their evaluated PNGs, and replays chart/CLEVR source programs.
The GQA source-program and image check is run by the separate reconstruction
command. A small machine may reduce `--workers`.

To run only the chart proof checker:

```bash
python scripts/verify_bounded_witness_bundle.py \
  --proofs evidence/plotqa/bounded-witness-proofs.jsonl.gz \
  --workers 8 --output ../chart-proof-verification.json
```

The prepared package includes historical counts plus current validation reports.
Reports are evidence of the checks they describe, not a guarantee about every
upstream annotation or a substitute for independent scientific review.