BlinkDL/RWKV-LM: RWKV is an RNN with transformer-level LLM performance. It can be directly trained like a GPT (parallelizable). So it's combining the best of RNN and transformer - great performance, fast inference, saves VRAM, fast training, "infinite" ctx_len, and free sentence embedding. - Cal's bookmarks

577 shaares

577 shaares

Filtres

Liens par page

20 50 100

BlinkDL/RWKV-LM: RWKV is an RNN with transformer-level LLM performance. It can be directly trained like a GPT (parallelizable). So it's combining the best of RNN and transformer - great performance, fast inference, saves VRAM, fast training, "infinite" ctx_len, and free sentence embedding.

https://news.ycombinator.com/item?id=35281026

ai

March 28, 2023 at 15:30:42 UTC * · permalien

·

·

https://github.com/BlinkDL/RWKV-LM

·

Filtres

Liens par page

20 50 100