tomaarsen / attention_sinks

Extend existing LLMs way beyond the original training length with constant memory usage, without retraining
https://huggingface.co/blog/tomaarsen/attention-sinks
Apache License 2.0
650 stars 41 forks source link

Add support for GPT-J models #11

Closed versae closed 9 months ago

versae commented 10 months ago

Any chance of adding support for GPT-J models?

tomaarsen commented 10 months ago

Great suggestion - will look into it.

versae commented 9 months ago

That was fast! 🙌🏼