# lightning.finite_horizon.studies.cuda_q1_posthoc { #aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc }
`lightning.finite_horizon.studies.cuda_q1_posthoc`
Admit the independently pinned CUDA C1 first-run observations.
This is a closed reader for ``cuda-q1-first-v1``, not an alternate generic
learning-cycle schema. It authenticates metadata and checkpoint bytes without
Torch deserialization, then validates historical fit/history and original-label
support. Runtime inference still uses the existing checkpoint verifier, which
checks actual tensor update buffers before loading weights.
## Classes
| Name | Description |
| --- | --- |
| [CudaQ1Posthoc](#aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.CudaQ1Posthoc) | Admitted fixed-weight development observations, never training authority. |
### CudaQ1Posthoc { #aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.CudaQ1Posthoc }
```python
lightning.finite_horizon.studies.cuda_q1_posthoc.CudaQ1Posthoc(
selected,
final,
train,
validation,
selected_rows,
final_rows,
history,
protocol,
config,
fit_receipt,
result,
provenance,
)
```
Admitted fixed-weight development observations, never training authority.
#### Attributes
| Name | Description |
| --- | --- |
| [selected](#aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.CudaQ1Posthoc.selected) | Historically validation-loss-selected secondary checkpoint. |
| [final](#aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.CudaQ1Posthoc.final) | Completed early-stopped final checkpoint; not a fixed-budget comparator. |
| [train](#aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.CudaQ1Posthoc.train) | Frozen ordered TRAIN routes; reopening still requires population admission. |
| [validation](#aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.CudaQ1Posthoc.validation) | Exposed development routes, disjoint from TRAIN physical scenes. |
| [selected_rows](#aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.CudaQ1Posthoc.selected_rows) | Original selected predictions paired by scene/store/rollout/state identity. |
| [final_rows](#aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.CudaQ1Posthoc.final_rows) | Original final predictions on identical candidate and original-label support. |
| [history](#aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.CudaQ1Posthoc.history) | Native validated terminal history, preserving native successful-update axes. |
| [protocol](#aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.CudaQ1Posthoc.protocol) | Frozen producer protocol, including source and population seals. |
| [config](#aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.CudaQ1Posthoc.config) | Original submitted configuration; execution overrides remain in fit receipt. |
| [fit_receipt](#aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.CudaQ1Posthoc.fit_receipt) | Authenticated executed scorer, module, trainer and dataset contracts. |
| [result](#aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.CudaQ1Posthoc.result) | Completed campaign terminal receipt, with original checkpoint summaries. |
| [provenance](#aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.CudaQ1Posthoc.provenance) | Independent source pins and limitations, separate from current inference. |
## Functions
| Name | Description |
| --- | --- |
| [read_cuda_q1_posthoc](#aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.read_cuda_q1_posthoc) | Admit S01 bytes and closed historical scientific identities. |
| [validate_cuda_q1_posthoc_evaluation](#aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.validate_cuda_q1_posthoc_evaluation) | Admit a fresh fixed-checkpoint evaluation against frozen CUDA support. |
### read_cuda_q1_posthoc { #aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.read_cuda_q1_posthoc }
```python
lightning.finite_horizon.studies.cuda_q1_posthoc.read_cuda_q1_posthoc(
root,
*,
source_pin,
)
```
Admit S01 bytes and closed historical scientific identities.
#### Parameters {.doc-section .doc-section-parameters}
| Name | Type | Description | Default |
|------------|-----------------------------------------------------|---------------------------------------------------------------------------------------------------------------------------------|------------|
| root | [Path](`pathlib.Path`) | Explicit local campaign root, without discovery or home expansion. | _required_ |
| source_pin | [dict](`dict`)\[[str](`str`), [Any](`typing.Any`)\] | Independently retained S01 entry from ``evidence-access.json``; never derive this trust input from the campaign being admitted. | _required_ |
#### Returns {.doc-section .doc-section-returns}
| Name | Type | Description |
|--------|--------------------------------------------------------------------------------------------|-----------------------------------------------------------------------------|
| | [CudaQ1Posthoc](`aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.CudaQ1Posthoc`) | Original paired development observations and sealed checkpoint references. |
| | [CudaQ1Posthoc](`aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.CudaQ1Posthoc`) | TRAIN routes are metadata-bound, not a claim of current store availability. |
#### Raises {.doc-section .doc-section-raises}
| Name | Type | Description |
|--------|----------------------------|------------------------------------------------------------------------------------------------------------------------------------|
| | [ValueError](`ValueError`) | If bytes, producer revision, native fit/history, population, successful updates, checkpoint role or original-label support differ. |
| | [OSError](`OSError`) | If required local source/checkpoint bytes cannot be read. |
### validate_cuda_q1_posthoc_evaluation { #aria_nbv.lightning.finite_horizon.studies.cuda_q1_posthoc.validate_cuda_q1_posthoc_evaluation }
```python
lightning.finite_horizon.studies.cuda_q1_posthoc.validate_cuda_q1_posthoc_evaluation(
admitted,
evaluation,
*,
split,
checkpoint,
)
```
Admit a fresh fixed-checkpoint evaluation against frozen CUDA support.
Full TRAIN uses the explicitly in-sample v2 receipt. Exposed development
validation retains v1. Predictions may differ with the recorded inference
device and evaluator implementation; original candidate identities, labels,
ordering and all population denominators must reproduce the historical
protocol exactly. No equality to old prediction-derived metrics is imposed.
#### Returns {.doc-section .doc-section-returns}
| Name | Type | Description |
|--------|-----------------------------------------------------------------------|-------------------------------------------------------------------------|
| | [list](`list`)\[[dict](`dict`)\[[str](`str`), [Any](`typing.Any`)\]\] | Complete native rows after checkpoint, population and numerical checks. |
#### Raises {.doc-section .doc-section-raises}
| Name | Type | Description |
|--------|----------------------------|-----------------------------------------------------------------------------------------------------------------------------|
| | [ValueError](`ValueError`) | If receipt roles, provenance, original support or native prediction-derived metrics disagree with their declared contracts. |