chore(docreader): 重新组织模块文件
This commit is contained in:
@@ -0,0 +1,21 @@
|
||||
from dataclasses import dataclass, field
|
||||
|
||||
|
||||
@dataclass
|
||||
class ChunkingConfig:
|
||||
"""
|
||||
Configuration for text chunking process.
|
||||
Controls how documents are split into smaller pieces for processing.
|
||||
"""
|
||||
|
||||
chunk_size: int = 512 # Maximum size of each chunk in tokens/chars
|
||||
chunk_overlap: int = 50 # Number of tokens/chars to overlap between chunks
|
||||
separators: list = field(
|
||||
default_factory=lambda: ["\n\n", "\n", "。"]
|
||||
) # Text separators in order of priority
|
||||
enable_multimodal: bool = (
|
||||
False # Whether to enable multimodal processing (text + images)
|
||||
)
|
||||
storage_config: dict = None # Preferred field name going forward
|
||||
vlm_config: dict = None # VLM configuration for image captioning
|
||||
|
||||
Reference in New Issue
Block a user