Currently we have a TokenCounter utility that will count tokens for drivers other than Ollama. Ollama currently doesn't provide token information in it's responses: https://github.com/ollama/ollama/issues/1716, but we could perhaps use a tokenizer to estimate token size?
This might be useful as well for customers to estimate a call before they run it.
Currently we have a TokenCounter utility that will count tokens for drivers other than Ollama. Ollama currently doesn't provide token information in it's responses: https://github.com/ollama/ollama/issues/1716, but we could perhaps use a tokenizer to estimate token size?
This might be useful as well for customers to estimate a call before they run it.