internlm

/

internlm3-8b-instruct

like 159

Follow

InternLM 383

Looks like 32k from config

1M, which is the same as InternLM2.5.

Thanks - does anything need changing in the config to allow it to use the extended context? Any other guidance? (I'm using TabbyAPI/exllamav2 but presumably anything that needs doing is applicable to others)

The usage for long context input is the same as for short context. You don't need to make any changes to the code, you can directly try inputting longer texts, and the model will automatically extrapolate.
If you're unsure about the input length, you can use len(tokenizer.encode(input)) to get the number of tokens in the input text.