Configuration Options¶

structx provides flexible configuration options for extraction.

Configuration Methods¶

You can configure structx in several ways:

YAML Configuration¶

# config.yaml
analysis:
  temperature: 0.2
  top_p: 0.1
  max_tokens: 2000

refinement:
  temperature: 0.1
  top_p: 0.05
  max_tokens: 2000

extraction:
  temperature: 0.0
  top_p: 0.1
  max_tokens: 2000
  frequency_penalty: 0.1

extractor = Extractor.from_litellm(
    model="gpt-4o-mini",
    api_key="your-api-key",
    config="config.yaml"
)

Dictionary Configuration¶

config = {
    "analysis": {
        "temperature": 0.2,
        "top_p": 0.1,
        "max_tokens": 2000
    },
    "refinement": {
        "temperature": 0.1,
        "top_p": 0.05,
        "max_tokens": 2000
    },
    "extraction": {
        "temperature": 0.0,
        "top_p": 0.1,
        "max_tokens": 2000,
        "frequency_penalty": 0.1
    }
}

extractor = Extractor.from_litellm(
    model="gpt-4o-mini",
    api_key="your-api-key",
    config=config
)

ExtractionConfig Object¶

from structx import ExtractionConfig, StepConfig

config = ExtractionConfig(
    analysis=StepConfig(
        temperature=0.2,
        top_p=0.1,
        max_tokens=2000
    ),
    refinement=StepConfig(
        temperature=0.1,
        top_p=0.05,
        max_tokens=2000
    ),
    extraction=StepConfig(
        temperature=0.0,
        top_p=0.1,
        max_tokens=2000,
        frequency_penalty=0.1
    )
)

extractor = Extractor.from_litellm(
    model="gpt-4o-mini",
    api_key="your-api-key",
    config=config
)

Configuration Parameters¶

Step Configuration¶

Each step in the extraction process can be configured separately:

Analysis: Query analysis to determine what to extract
Refinement: Query refinement and model generation
Extraction: Actual data extraction

Common Parameters¶

Parameter	Type	Default	Description
temperature	float	varies	Sampling temperature (0.0-1.0)
top_p	float	varies	Nucleus sampling parameter (0.0-1.0)
max_tokens	int	2000	Maximum tokens in completion

Default Values¶

Step	Temperature	Top P	Max Tokens
Analysis	0.2	0.1	2000
Refinement	0.1	0.05	2000
Extraction	0.0	0.1	2000

Retry Configuration¶

You can configure the retry behavior for extraction:

extractor = Extractor.from_litellm(
    model="gpt-4o-mini",
    api_key="your-api-key",
    max_retries=5,      # Maximum number of retry attempts
    min_wait=2,         # Minimum seconds to wait between retries
    max_wait=30         # Maximum seconds to wait between retries
)

Retry Parameters¶

Parameter	Type	Default	Description
max_retries	int	3	Maximum number of retry attempts
min_wait	int	1	Minimum seconds to wait between retries
max_wait	int	10	Maximum seconds to wait between retries

Processing Configuration¶

You can configure the processing behavior:

extractor = Extractor.from_litellm(
    model="gpt-4o-mini",
    api_key="your-api-key",
    max_threads=20,     # Maximum number of concurrent threads
    batch_size=50       # Size of processing batches
)

Processing Parameters¶

Parameter	Type	Default	Description
max_threads	int	10	Maximum number of concurrent threads
batch_size	int	100	Size of processing batches

Best Practices¶

Temperature Settings:
Use lower temperatures (0.0-0.2) for consistent extraction
Higher temperatures may introduce variability
Token Limits:
Ensure max_tokens is sufficient for your extraction needs
Complex extractions may require higher limits
Batch Size:
Adjust based on your data size and memory constraints
Smaller batches use less memory but may be slower
Thread Count:
Set based on your CPU capabilities
Too many threads can cause resource contention