nemo_gym.harness_capabilities.results

View as Markdown

Independent check results. TE labels describe results; they never schedule checks.

Module Contents

Classes

NameDescription
CheckResult-
ResultsCombine repeated evaluations of one rule without hiding failures or blocked checks.

Functions

NameDescription
gate_passesRequire all applicable checks in the selected tier, independently of TE labels.
render_matricesRender saved check results; never infer a pass from an absent failure message.

Data

Status

API

class nemo_gym.harness_capabilities.results.CheckResult(
id: str,
evidence: tuple[str, ...] = (),
locations: list[str] = list(),
reasons: list[str] = list(),
blocked_by: list[str] = list()
)
Dataclass
blocked_by
list[str] = field(default_factory=list)
evidence
tuple[str, ...] = ()
id
str
kind
Kind
locations
list[str] = field(default_factory=list)
reasons
list[str] = field(default_factory=list)
status
Status
tier
PriorityTier
class nemo_gym.harness_capabilities.results.Results()

Combine repeated evaluations of one rule without hiding failures or blocked checks.

rows
dict[str, CheckResult] = {}
nemo_gym.harness_capabilities.results.Results.add(
check_id: str,
evidence: tuple[str, ...] = (),
location: str = '',
reason: str = '',
blocked_by: collections.abc.Iterable[str] = ()
) -> bool
nemo_gym.harness_capabilities.results.Results.dump() -> list[dict]
nemo_gym.harness_capabilities.results.Results.run(
) -> bool

Execute a typed rule; excluded and blocked rules never evaluate their input.

nemo_gym.harness_capabilities.results.gate_passes(
checks: list[dict],
) -> bool

Require all applicable checks in the selected tier, independently of TE labels.

nemo_gym.harness_capabilities.results.render_matrices(
harnesses: dict
) -> str

Render saved check results; never infer a pass from an absent failure message.

nemo_gym.harness_capabilities.results.Status = Literal['pass', 'fail', 'not_applicable']