What is the context size on this model? And it does not appear to deal with JSON, function calling well.
#15
by
BigDeeper
- opened
Has it been trained for function calling?
Context length appears to be 32K tokens according to config.json
Tokenizer config doesn't feature the special function calling tokens though, so I am pretty sure it wasn't fine-tuned for that.
That said, there might be a mishap with the tokenizer config, there's an ongoing discussion about it. I am not sure about that.