Examples
Comprehensive examples demonstrating Kerb’s capabilities across all modules.
By Module
Generation
Text generation examples with multiple providers.
01_basic_generation.py - Basic text generation
02_streaming_generation.py - Streaming responses
03_batch_generation.py - Batch processing
04_multi_provider_comparison.py - Compare providers
05_cost_tracking.py - Track API costs
06_structured_output.py - Structured outputs
07_retry_and_error_handling.py - Error handling
08_rate_limiting.py - Rate limiting
09_response_caching.py - Response caching
10_model_switching.py - Dynamic model switching
Context Management
Context window and token management examples.
01_basic_context_window.py - Basic context windows
02_truncation_strategies.py - Truncation strategies
03_priority_management.py - Priority management
04_context_compression.py - Context compression
05_sliding_windows.py - Sliding window patterns
06_context_optimization.py - Context optimization
07_context_formatting.py - Context formatting
08_conversational_context.py - Conversational contexts
09_rag_context.py - RAG context management
Parsing
Output parsing and validation examples.
01_json_extraction.py - JSON extraction
02_pydantic_models.py - Pydantic validation
03_function_calling.py - Function calling
04_code_extraction.py - Code extraction
05_text_extraction.py - Text extraction
06_robust_parsing.py - Robust parsing
07_validation.py - Schema validation
08_extraction_pipeline.py - Extraction pipelines
Preprocessing
Text preprocessing and cleaning examples.
01_text_normalization.py - Text normalization
02_content_filtering.py - Content filtering
03_deduplication.py - Deduplication
04_language_detection.py - Language detection
05_batch_processing.py - Batch processing
06_content_analysis.py - Content analysis
07_transformations.py - Text transformations
08_production_pipelines.py - Production pipelines
Evaluation
Model evaluation and benchmarking examples.
01_ground_truth_metrics.py - Ground truth metrics
02_quality_assessment.py - Quality assessment
03_llm_as_judge.py - LLM-as-a-judge
04_benchmarking.py - Benchmarking
05_ab_testing.py - A/B testing
06_rag_evaluation.py - RAG evaluation
07_model_comparison.py - Model comparison
Prompt Engineering
Prompt templates and optimization examples.
01_template_basics.py - Template basics
04_optimization.py - Prompt optimization
05_advanced_patterns.py - Advanced patterns
Retrieval
RAG and semantic search examples.
04_context_management.py - Context management
05_rag_pipeline.py - RAG pipeline
06_multi_query_retrieval.py - Multi-query retrieval
08_result_formatting.py - Result formatting
Fine-Tuning
Model fine-tuning and dataset preparation examples.
See Fine-Tuning Documentation for detailed guides:
Quick Start Examples
Basic Generation
from kerb.generation import generate, ModelName, LLMProvider
response = generate(
"Explain quantum computing",
model=ModelName.GPT_4O_MINI,
provider=LLMProvider.OPENAI
)
print(response.content)
RAG Pipeline
from kerb.document import load_document
from kerb.chunk import chunk_text
from kerb.embedding import embed, embed_batch
from kerb.retrieval import semantic_search, Document
from kerb.generation import generate, ModelName
# Load and process
doc = load_document("paper.pdf")
chunks = chunk_text(doc.content, chunk_size=512)
embeddings = embed_batch(chunks)
# Search
query_embedding = embed("What are the findings?")
documents = [Document(content=c) for c in chunks]
results = semantic_search(query_embedding, documents, embeddings, top_k=5)
# Generate
context = "\n".join([r.document.content for r in results])
answer = generate(f"Based on: {context}\n\nQuestion: What are the findings?")
Response Caching
from kerb.cache import create_memory_cache, generate_prompt_key
from kerb.generation import generate, ModelName
cache = create_memory_cache(max_size=1000)
def cached_generate(prompt, model=ModelName.GPT_4O_MINI):
key = generate_prompt_key(prompt, model=model.value)
if cached := cache.get(key):
return cached['response']
response = generate(prompt, model=model)
cache.set(key, {'response': response})
return response
Running Examples
All examples are self-contained and can be run directly:
# Set your API keys
export OPENAI_API_KEY="your-key"
export ANTHROPIC_API_KEY="your-key"
# Run any example
python docs/examples/generation/01_basic_generation.py
# Or use the run script
python scripts/run_examples.py --module generation
Additional Resources
API Reference - Complete API documentation
Getting Started - Quick start guide
GitHub Examples - All example code