Generation Configuration
- class DiagramGenerationConfig(**data)
Bases:
BaseModelCreate a new model by parsing and validating input data from keyword arguments.
Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.
self is explicitly positional-only to allow self as a field name.
- min_pts: int
The minimum number of points added to the original diagram.
- max_pts: int
The maximum number of points added to the original diagram.
- initial_jgex_problem: Optional[str]
If the user provides this, we’ll use the JGEX problem as a starting point for the generation.
- output_gcs_file: str | None
- emit_only_double_checked_aux_subproblems: bool
If true, we only emit subproblems that require aux constructions.
- aux_discriminator: Literal['he_point_deps', 'newclid_traceback']
The discriminator to use to split subgoals points into setup and aux points.
- debug: bool
Whether to run in single-process debug mode.
- debug_problem: Optional[str]
If the user provides this, we’ll use it as the problem.
- double_check_probability: float
Probability of double-checking the subproblem with and without auxiliary constructions. Will only apply to subproblems with at least one auxiliary constructions.
- skip_aux_setup_pairs_previous_false_positive: bool
If true, we will skip subproblems that have an aux setup pair that was a false positive on a previous goal.
- min_rules_applied: int
The minimum number of rules applied in the subproblem to not be skipped.
- random_seed: Optional[int]
If provided, we will use this seed for the random number generator. Otherwise, we will use the current time as a seed.
- attempts_per_diagram_build: int
How many times we try to find a configuration of points in 2D that satisfy the assumptions of the problem.
- max_check_per_predicate: int
The maximum number of times we will double-check a subproblem with goal of each predicate type within a diagram.
- construction_counter_reporting_rate: float
How often we report the construction attempt counters. Should be between 0 and 1.
- attempts_per_clause: int
Each time we sample a JGEX clause, we will try to find input arguments that work this number of times before giving up on it. The input arguments are sampled uniformly at random from the set of all possible input arguments.
- aux_tag: Literal['!aux', '']
The tag to use for prefixing auxiliary constructions.
- pmf_additional_free_points_sweep: list[dict[int, float]]
- model_config: ClassVar[ConfigDict] = {}
Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].
- pmf_intersect_vs_other_sweep: list[dict[IntersectCases, float]]
- pmf_num_intersecting_to_sample_sweep: list[dict[int, float]]
- run_uuid: str
- timeout: float | None
Timeout (in seconds) for the generation process. (Used in tests)
- n_workers: int | Literal['auto']
- jsonl_dump_file: Path
- log_level: Literal['DEBUG', 'INFO', 'WARNING', 'ERROR', 'CRITICAL']
The log level to use for the generation process.
- classmethod validate_pmf_sweep(pmf_sweep)
- Return type:
list[dict[TypeVar(T),float]]
- classmethod validate_output_gcs_file(gcs_file)
- Return type:
str|None
- validate_min_max_pts()
- Return type:
Self
- class SweepSelection(**data)
Bases:
BaseModelCreate a new model by parsing and validating input data from keyword arguments.
Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.
self is explicitly positional-only to allow self as a field name.
- model_config: ClassVar[ConfigDict] = {}
Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].
- additional_free_points_sweep_selection: int
- intersect_vs_other_sweep_selection: int
- num_intersecting_to_sample_sweep_selection: int
- sample_sweep_parameters(cfg, rng)
Sample sweep parameters from the configuration and return a new config instance.
- Parameters:
cfg (
DiagramGenerationConfig) – The base configuration to sample from- Return type:
- Returns:
A new configuration instance with sampled sweep parameters