Closed MrWangGang closed 1 year ago
gpt-3.5-turbo uses cl100k_base as encoding, but you can simply use getEncondingForModel:
EncodingRegistry registry = Encodings.newDefaultEncodingRegistry();
Encoding enc = registry.getEncodingForModel(ModelType.GPT_3_5_TURBO);
For further information on how to use Encoding, refer to the docs: https://javadoc.io/doc/com.knuddels/jtokkit/latest/index.html
If you are talking client libraries: this is out of scope for this library, but if you need a convenient API client you can take a lookt at https://github.com/TheoKanning/openai-java
Thank you for your answer. It's very useful to me
May I ask if you support the GPT-3.5-tubor model, and if so, how can I use it?