Hi 馃憢馃徎, first of all, I'm loving this project! Keep up the great work on this awesome project.
Anthropic recently released Claude 3.7 Sonnet, which supports an output of up to 128K tokens. I鈥檇 love to use this extended output in srcbook. What changes would be needed to support this?
I made a naive solution by updating the generateText call to set maxTokens: 128000 and modifying the createAnthropic call to add the anthropic-beta header when using claude-3.7-sonnet models.
Would this be the right approach, or are there other places that need updates?
I think the best solution would be a configuration for max tokens and maybe a check for values > 8192 (the original maxTokens).
Happy to help with implementation if needed! 馃殌
Here is the list of Anthropic models.
And this is the info for the extended output header.
Hi 馃憢馃徎, first of all, I'm loving this project! Keep up the great work on this awesome project.
Anthropic recently released Claude 3.7 Sonnet, which supports an output of up to 128K tokens. I鈥檇 love to use this extended output in srcbook. What changes would be needed to support this?
I made a naive solution by updating the
generateTextcall to setmaxTokens: 128000and modifying thecreateAnthropiccall to add theanthropic-betaheader when using claude-3.7-sonnet models.Would this be the right approach, or are there other places that need updates?
I think the best solution would be a configuration for max tokens and maybe a check for values > 8192 (the original
maxTokens).Happy to help with implementation if needed! 馃殌
Here is the list of Anthropic models.
And this is the info for the extended output header.
