Full version of H3 model?

[FeelingFatigued]

Hello,

This repository does not seem to have the complete H3 described in the paper.
(e.g., sequence length must be <= 8192 and I couldn't find state passing algorithm in the source)

If it is possible, would you please push the most recent version of the H3 model described in the paper?

[DanFu09]

We have a new kernel coming out very soon that obviates that algorithm and scales up to 4M - will post here when we release it!

[FeelingFatigued]

I noticed the new FlashFFTConv repo was released recently. I've been trying to use that in this H3 model but I am not sure if I am changing the code correctly.

If it is possible, could you update H3 repo with FlashFFTConv? or could you tell me where do I need to change?

Thanks in advnace.

[DanFu09]

Ack’ing that I’ve seen this, will try to get it in this week or next!

…

On Fri, Nov 24, 2023 at 3:07 AM FeelingFatigued ***@***.***> wrote: I noticed the new FlashFFTConv repo was released recently. I've been trying to use that in this H3 model but I am not sure if I am changing the code correctly. If it is possible, could you update H3 repo with FlashFFTConv? or could you tell me where do I need to change? Thanks in advnace. — Reply to this email directly, view it on GitHub <#29 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABDDIIUOZSRQUODR7NVUOUTYGBIUBAVCNFSM6AAAAAA6VRHSM2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRVGI4DKMRXGU> . You are receiving this because you commented.Message ID: ***@***.***>

[FeelingFatigued]

Sorry but can I know when will the FlashFFTConv-enabled H3 be available?

Thanks for your help.