Evidence

Compare two experiments

Check what changed in the question, wording or readers before comparing the numbers.

This view keeps both results separate. It does not calculate a combined score or decide whether they reproduce each other.

Change the selected experimentsSearch by proposal, select a result or enter an exact identifier

Find experiments by proposal

Choose two experiments by title, measurement and date. The choices include up to 50 newest public completed results, plus your current selections. Historical results stay labelled. For older records, use the evidence explorer or exact entry below.

Use an exact experiment identifier instead

A full attempt UUID entered here replaces the corresponding choice above. Content hashes are not result identifiers.

Clear comparison

Compare the essentials

Read each question across both results. A matching description is not proof that the studies are scientifically comparable.

Question	First result	Second result
Which proposal?	resume-from / redo-from-start — does earlier work still count?2026-09-18 08:31 UTC	resume-from / redo-from-start — does earlier work still count?2026-09-18 16:59 UTC
What was measured?	comprehension accuracyHow does the wording change correct answers from the declared reader panel?	comprehension accuracyHow does the wording change correct answers from the declared reader panel?
What was the result?	-7.205 percentage pointsReported interval: -25.2389 to 10.2627Not yet counting in evidence decisions	-6.025 percentage pointsReported interval: -23.3586 to 12.0547Counts in current evidence decisions
Compared with what?	Other declared comparison; inspect the specificationReader exposure not recorded as a structured label	Other declared comparison; inspect the specificationReader exposure not recorded as a structured label
Independently settled?	DisputedEligible replications disagree and this original does not hold a settlement majority.	Disagrees with the named originalThis eligible row adds one disagreement. An adverse or null direction is a valid result and remains visible.

Inspect first result Inspect second result

What differs in the declared methods?

These are literal recorded values, not a compatibility test. Matching declarations do not prove equivalent inputs or fair scoring. A missing structured field may be described elsewhere in the immutable specification.

What was measured · Same recorded value

First result

comprehension_accuracy_delta

Second result

comprehension_accuracy_delta

English comparison declarations · Same recorded value

First result

[
    "complete-canonical-concise-english-v1"
]

Second result

[
    "complete-canonical-concise-english-v1"
]

Tested population · Not recorded on either side

First result

Not recorded in this structured field

Second result

Not recorded in this structured field

Unit tested · Not recorded on either side

First result

Not recorded in this structured field

Second result

Not recorded in this structured field

Named instruments (including order) · Same recorded value

First result

[
    "Saturnia-Verified-Gemma12@q4_k_m",
    "Saturnia-Verified-Mistral24@q4_k_m"
]

Second result

[
    "Saturnia-Verified-Gemma12@q4_k_m",
    "Saturnia-Verified-Mistral24@q4_k_m"
]

Reader exposure · Not recorded on either side

First result

Not recorded in this structured field

Second result

Not recorded in this structured field

Exposure window · Not recorded on either side

First result

Not recorded in this structured field

Second result

Not recorded in this structured field

How results combine · Not recorded on either side

First result

Not recorded in this structured field

Second result

Not recorded in this structured field

Scoring declaration · Not recorded on either side

First result

Not recorded in this structured field

Second result

Not recorded in this structured field

Conditions (including order) · Same recorded value

First result

[
    {
        "id": "resume-core",
        "weight": 1
    },
    {
        "id": "redo-core",
        "weight": 1
    }
]

Second result

[
    {
        "id": "resume-core",
        "weight": 1
    },
    {
        "id": "redo-core",
        "weight": 1
    }
]

First result: full recordcomprehension accuracy · -7.205 percentage points

First result · 2026-09-18 08:31 UTC

resume-from / redo-from-start — does earlier work still count?

Not yet counting in evidence decisions. This row remains available for assessment, but does not currently carry a counting evidence result.

What was measured

comprehension accuracy · comprehension_accuracy_delta
How does the wording change correct answers from the declared reader panel?

Reported result

-7.205 percentage points
Reported interval: -25.2389 to 10.2627.

The value is neutral or does not resolve the registered direction.

English comparison

Other declared comparison; inspect the specification

Declared by the submitter; not a certification of equivalent information.

Exact registered concise templates; task and checkpoint facts unchanged in both arms. Core-only scope.

Reader exposure

Reader exposure not recorded as a structured label

Named instruments

Saturnia-Verified-Gemma12@q4_k_m, Saturnia-Verified-Mistral24@q4_k_m

Reader population not separately declared.

Conditions covered

Separate outcomes retained for all 2 declared conditions

resume-core, redo-core

An overall average can hide a weak condition. A condition list is not proof that every form or claim in the proposal was tested.

Settlement role

Disputed

Eligible replications disagree and this original does not hold a settlement majority.

How often did each version lead to the right answer?

English comparison: 57.99%; 57.99%
Ainglish version: 50.78%; 50.78%

Reported real-item accuracy, not the separate calibration score. Both bars use the same 0–100% scale. The difference is measured in percentage points, not percent improvement. Any declared stratum weights are already applied.

Reported item-bootstrap interval: -25.2389 to 10.2627 percentage points.

This interval concerns the difference, not separate uncertainty bounds for either accuracy. It does not measure uncertainty across humans or future models.

At least one declared condition is resolution-limited. The overall interval does not settle every condition.

Real cases: 64 · Named readers: 2. These are different units; multiple answers to one case are not new cases.

Does the overall result hide differences between conditions?

Every stored condition, without new pooling. Differences and intervals use percentage points. Condition names come from the frozen experiment.
Condition	Reported difference	Reported interval	English accuracy	Ainglish accuracy
`resume-core`	-6.2	Not recorded	43.24%	37.04%
`redo-core`	-8.21	Not recorded	72.73%	64.52%

A missing condition interval is not zero uncertainty. An overall interval cannot substitute for agreement in every load-bearing condition.

Inspect externally stored inputs and recorded answers

The comparison label is the submitter’s declaration, not a semantic certification. Check that both versions preserve the information needed to answer the same question.

Numbers count only readable inputs attached to this receipt. They are not the experiment’s declared sample size or the number of reader calls.

The input material is linked externally. The number of study items and controls in that file has not been checked by this website. “External file” does not mean zero inputs.

Open the declared external input artifact. This is an unverified external link, not a hosted or inspected copy.

Declared input digest: 0a7cf3c482bd5b30ab37c61c7f253d655e40200ebf00ae87475b691de9aef8cc. A recorded digest alone does not establish that the linked file matches it.

The website does not fetch the file. Verify the declared digest recipe before relying on it: SDK item digests use canonical JSON of the item array, not the raw pretty-printed file bytes.

No readable study input pairs are stored inline in this receipt. This does not mean the experiment used none.

Prompts, reference material and other context can live elsewhere in the specification. Inputs and keys alone do not reconstruct every reader call or establish a fair comparison.

Declared population, method and retained outcomes

No structured study scope is declared here. Inspect the immutable manifest; do not infer a comparator or population from the headline.

Absolute arm results, reader-specific results and condition results below are retained values, not a newly pooled analysis. Accuracy arms use fractions from 0 to 1; their difference uses percentage points.

Absolute arm results

{
    "english": 0.57989999999999997104538351777591742575168609619140625,
    "ainglish": 0.507800000000000029132252166164107620716094970703125,
    "chance": 0.5
}

Reader or tokenizer results

[
    {
        "model": "Saturnia-Verified-Gemma12",
        "value": -28.9200000000000017053025658242404460906982421875,
        "precision": "q4_k_m"
    },
    {
        "model": "Saturnia-Verified-Mistral24",
        "value": 15.8800000000000007815970093361102044582366943359375,
        "precision": "q4_k_m"
    }
]

Condition results

[
    {
        "id": "resume-core",
        "weight": 1,
        "share": 0.5,
        "value": -6.20000000000000017763568394002504646778106689453125,
        "value_lo": null,
        "value_hi": null,
        "arms": {
            "english": 0.43240000000000000657252030578092671930789947509765625,
            "ainglish": 0.370400000000000007016609515630989335477352142333984375,
            "chance": 0.5
        },
        "resolution_bound": "floor"
    },
    {
        "id": "redo-core",
        "weight": 1,
        "share": 0.5,
        "value": -8.21000000000000085265128291212022304534912109375,
        "value_lo": null,
        "value_hi": null,
        "arms": {
            "english": 0.72729999999999994653165913405246101319789886474609375,
            "ainglish": 0.64519999999999999573674358543939888477325439453125,
            "chance": 0.5
        },
        "resolution_bound": "resolvable"
    }
]

Exact result and immutable specification Experiment history

Attempt b6260fc3-cab8-406d-81ea-2629fe935048
Content 763f2a4163f3813f863c8f33e7ec11f77bd5c8c74bffd16b657f24e037f46514

Second result: full recordcomprehension accuracy · -6.025 percentage points

Second result · 2026-09-18 16:59 UTC

resume-from / redo-from-start — does earlier work still count?

Counts in current evidence decisions. This row currently contributes to evidence decisions. Its direction is separate from whether the proposal is ready for adoption.

What was measured

comprehension accuracy · comprehension_accuracy_delta
How does the wording change correct answers from the declared reader panel?

Reported result

-6.025 percentage points
Reported interval: -23.3586 to 12.0547.

The value is neutral or does not resolve the registered direction.

English comparison

Other declared comparison; inspect the specification

Declared by the submitter; not a certification of equivalent information.

Exact registered concise templates; task and checkpoint facts unchanged in both arms. Core-only scope.

Reader exposure

Reader exposure not recorded as a structured label

Named instruments

Saturnia-Verified-Gemma12@q4_k_m, Saturnia-Verified-Mistral24@q4_k_m

Reader population not separately declared.

Conditions covered

Separate outcomes retained for all 2 declared conditions

resume-core, redo-core

An overall average can hide a weak condition. A condition list is not proof that every form or claim in the proposal was tested.

Settlement role

Disagrees with the named original

This eligible row adds one disagreement. An adverse or null direction is a valid result and remains visible.

This result checks a named original, not every experiment on the proposal. Read its target original

Compare with the exact target attempt

Every declared condition must agree. Overlapping overall intervals alone do not confirm this original.

How much input text was reused?

Complete-pair freshness is not available for this receipt.

Separate-arm overlap is unavailable or has not been computed. This does not mean zero reuse.

The item banks are referenced by digest rather than inspectable here. Compare the explicitly retrieved, digest-verified files before making a freshness claim.

Declared item-bank digests: different. This compares bank identity, not shared sentences; different bank digests can still contain identical pairs.

Exact text comparisons only; repeated occurrences count separately. Shared text can deserve scrutiny even when each complete pair is new. These arm counts are descriptive and do not change settlement eligibility.

Declared target content identity

763f2a4163f3813f863c8f33e7ec11f77bd5c8c74bffd16b657f24e037f46514

How often did each version lead to the right answer?

English comparison: 53.35%; 53.35%
Ainglish version: 47.32%; 47.32%

Reported item-bootstrap interval: -23.3586 to 12.0547 percentage points.

This interval concerns the difference, not separate uncertainty bounds for either accuracy. It does not measure uncertainty across humans or future models.

At least one declared condition is resolution-limited. The overall interval does not settle every condition.

Real cases: 64 · Named readers: 2. These are different units; multiple answers to one case are not new cases.

Does the overall result hide differences between conditions?

Every stored condition, without new pooling. Differences and intervals use percentage points. Condition names come from the frozen experiment.
Condition	Reported difference	Reported interval	English accuracy	Ainglish accuracy
`resume-core`	-5.77	Not recorded	45.16%	39.39%
`redo-core`	-6.28	Not recorded	61.54%	55.26%

A missing condition interval is not zero uncertainty. An overall interval cannot substitute for agreement in every load-bearing condition.

Inspect externally stored inputs and recorded answers

The comparison label is the submitter’s declaration, not a semantic certification. Check that both versions preserve the information needed to answer the same question.

Numbers count only readable inputs attached to this receipt. They are not the experiment’s declared sample size or the number of reader calls.

The input material is linked externally. The number of study items and controls in that file has not been checked by this website. “External file” does not mean zero inputs.

Open the declared external input artifact. This is an unverified external link, not a hosted or inspected copy.

Declared input digest: d2ee5b418a037e1054ccb1691bcf4c721ea2332c51ff2de6ae196bcb267ed940. A recorded digest alone does not establish that the linked file matches it.

The website does not fetch the file. Verify the declared digest recipe before relying on it: SDK item digests use canonical JSON of the item array, not the raw pretty-printed file bytes.

No readable study input pairs are stored inline in this receipt. This does not mean the experiment used none.

Prompts, reference material and other context can live elsewhere in the specification. Inputs and keys alone do not reconstruct every reader call or establish a fair comparison.

Declared population, method and retained outcomes

No structured study scope is declared here. Inspect the immutable manifest; do not infer a comparator or population from the headline.

Absolute arm results

{
    "english": 0.53349999999999997424282582869636826217174530029296875,
    "ainglish": 0.473200000000000009503509090791339986026287078857421875,
    "chance": 0.5
}

Reader or tokenizer results

[
    {
        "model": "Saturnia-Verified-Gemma12",
        "value": 8.44500000000000028421709430404007434844970703125,
        "precision": "q4_k_m"
    },
    {
        "model": "Saturnia-Verified-Mistral24",
        "value": -20.449999999999999289457264239899814128875732421875,
        "precision": "q4_k_m"
    }
]

Condition results

[
    {
        "id": "resume-core",
        "weight": 1,
        "share": 0.5,
        "value": -5.769999999999999573674358543939888477325439453125,
        "value_lo": null,
        "value_hi": null,
        "arms": {
            "english": 0.45160000000000000142108547152020037174224853515625,
            "ainglish": 0.393899999999999972377651147326105274260044097900390625,
            "chance": 0.5
        },
        "resolution_bound": "floor"
    },
    {
        "id": "redo-core",
        "weight": 1,
        "share": 0.5,
        "value": -6.28000000000000024868995751603506505489349365234375,
        "value_lo": null,
        "value_hi": null,
        "arms": {
            "english": 0.6153999999999999470645661858725361526012420654296875,
            "ainglish": 0.5525999999999999801048033987171947956085205078125,
            "chance": 0.5
        },
        "resolution_bound": "resolvable"
    }
]

Exact result and immutable specification Experiment history

Attempt 2ee664b9-b95c-47d3-be96-04d75b01078d
Content 3e86ce2bf89d5fc63044f8eec8fcf9189558354afabe55355f166fe63d3e7f75

Different wording, readers, exposure or populations can legitimately produce different results. A visible reference is not training the model’s weights. Current models and tokenizers have learned English; future Ainglish-trained performance remains a research question.