Spaces:
Sleeping
Sleeping
File size: 3,467 Bytes
5f45585 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 |
from pydantic import BaseModel, Field
from typing import Any, Dict, List, Optional
class Action(BaseModel):
"""Action that the agent can take in the environment."""
operation: str = Field(
...,
description=(
"Operation to perform. One of: "
"remove_duplicates, fill_missing, fix_dtype, "
"remove_outliers, rename_columns, validate_schema, finish"
)
)
parameters: Dict[str, Any] = Field(
default_factory=dict,
description="Parameters for the operation"
)
class Config:
json_schema_extra = {
"example": {
"operation": "fill_missing",
"parameters": {
"column": "age",
"strategy": "mean"
}
}
}
class Observation(BaseModel):
"""What the agent sees at each step."""
task_id: str = Field(..., description="Current task identifier")
step: int = Field(..., description="Current step number")
dataset_info: Dict[str, Any] = Field(
...,
description="Info about current dataset state"
)
columns: List[str] = Field(
...,
description="Column names in dataset"
)
shape: List[int] = Field(
...,
description="[rows, columns] of dataset"
)
missing_values: Dict[str, int] = Field(
...,
description="Count of missing values per column"
)
dtypes: Dict[str, str] = Field(
...,
description="Data types of each column"
)
duplicate_count: int = Field(
...,
description="Number of duplicate rows"
)
sample_rows: List[Dict[str, Any]] = Field(
...,
description="First 3 rows of dataset as preview"
)
available_operations: List[str] = Field(
...,
description="List of valid operations agent can use"
)
task_description: str = Field(
...,
description="Natural language description of what agent must do"
)
message: str = Field(
default="",
description="Feedback message from last action"
)
class Reward(BaseModel):
"""Reward signal returned after each step."""
total: float = Field(
...,
description="Total reward this step (0.0 to 1.0)"
)
duplicate_score: float = Field(
default=0.0,
description="Score for duplicate removal"
)
missing_score: float = Field(
default=0.0,
description="Score for handling missing values"
)
dtype_score: float = Field(
default=0.0,
description="Score for fixing data types"
)
outlier_score: float = Field(
default=0.0,
description="Score for outlier removal"
)
schema_score: float = Field(
default=0.0,
description="Score for schema validation"
)
penalty: float = Field(
default=0.0,
description="Penalty for bad actions"
)
class StepResult(BaseModel):
"""Full result returned by step()."""
observation: Observation
reward: Reward
done: bool = Field(..., description="Whether episode is complete")
info: Dict[str, Any] = Field(
default_factory=dict,
description="Extra info for debugging"
)
class TaskInfo(BaseModel):
"""Metadata about a task."""
task_id: str
difficulty: str
description: str
max_steps: int
operations_allowed: List[str]
|