38 1 31

BlinkDL

https://www.rwkv.com/

AI & ML interests

RWKV is all you need

Recent Activity

updated a Space 13 minutes ago

BlinkDL/RWKV-Gradio-1

updated a model about 8 hours ago

BlinkDL/rwkv-7-world

updated a model 4 days ago

BlinkDL/temp-latest-training-models

View all activity

Organizations

BlinkDL's activity

posted an update 21 days ago

Post

1985

RWKV-7 "Goose" 0.4B trained w/ ctx4k automatically extrapolates to ctx32k+, and perfectly solves NIAH ctx16k 🤯 100% RNN and attention-free. Only trained on the Pile. No finetuning. Replicable training runs. tested by our community: https://github.com/Jellyfish042/LongMamba

posted an update about 2 months ago

Post

4505

RWKV-6-world-v3 (+3.1T tokens) is our best multilingual 7B model as of now: BlinkDL/rwkv-6-world

It's 100% RNN and attention-free. MMLU 54.2% (previous world-v2.1 = 47.9%. note: without eval-boosting tricks such as annealing).

RWKV-7-world-v4 soon :)

posted an update 4 months ago

Post

5477

RWKV-7 "Goose" preview rc2 => Peak RNN architecture?😃Will try to squeeze more performance for the final release. Preview code & model: https://github.com/BlinkDL/RWKV-LM/tree/main/RWKV-v7

2 replies