• Explain the need for batching due to embedding model thresholds
  • Describe the BatchingStrategy interface and its purpose
  • Detail the default TokenCountBatchingStrategy implementation
  • Provide information on using and customizing batching strategies
  • Mention that several vector stores are pre-configured with the default strategy

Comment From: markpollack

Added new constructor for custom TokenCountEstimator in TokenCountBatchingStrategy and null checks in ctors

updated doc with example usage of additional constructors

merged in 4c8a6ee8b0deec797901f1164cb63dc5340f0b84