Lower default chunking max segment sizes

[melw66]

For each embedding model, lowered the default chunking max segment size in characters from the max character limit to a lower value for each Bedrock embedding model so that model token count limits are not hit with the default configuration.

• Amazon Titan Text Embeddings V1
  • Token limit: 8,192 tokens
  • Max character limit: 50,000
  • New default character limit: 16,000
• Amazon Titan Text Embeddings V2
  • Token limit: 8,192 tokens
  • Max character limit: 50,000
  • New default character limit: 16,000
• Amazon Titan Multimodal Embeddings G1
  • Token limit: 128 tokens
  • Max character limit: 100,000
  • New default character limit: 50,000
• Cohere Embed English
  • Token limit: 512 tokens
  • Max character limit: 2048
  • New default character limit: 1000
• Cohere Embed Multilingual
  • Token limit: 512 tokens
  • Max character limit: 2048
  • New default character limit: 1000