BlinkDL (@blinkdl_ai) 's Twitter Profile
BlinkDL

@blinkdl_ai

RWKV = 100% RNN with GPT-level performance. lfaidata.foundation/projects/rwkv and github.com/search?o=desc&…

ID: 1570825439177998336

linkhttps://www.rwkv.com/ calendar_today16-09-2022 17:22:39

358 Tweet

8,8K Followers

148 Following

BlinkDL (@blinkdl_ai) 's Twitter Profile Photo

Falcon-H1 is apparently eval-maxxing, but better than expected at uncheatable_eval, so it's a decent base model 🙂 all results: github.com/Jellyfish042/u…

Falcon-H1 is apparently eval-maxxing, but better than expected at uncheatable_eval, so it's a decent base model 🙂 all results: github.com/Jellyfish042/u…
BlinkDL (@blinkdl_ai) 's Twitter Profile Photo

RWKV papers rwkv.com : 15 new in Apr/May 2025 🔥 DualComp using RWKV-7 for efficient compression, and RWKVQuant doing 3.275bit. RWKV-7 "Goose" 🪿 is 100% RNN and efficiently test-time-training its state via in-context gradient descent at every token in parallel.

RWKV papers rwkv.com : 15 new in Apr/May 2025 🔥 DualComp using RWKV-7 for efficient compression, and RWKVQuant doing 3.275bit. RWKV-7 "Goose" 🪿 is 100% RNN and efficiently test-time-training its state via in-context gradient descent at every token in parallel.
BlinkDL (@blinkdl_ai) 's Twitter Profile Photo

Songlin blocked me on X and banned me from FLA discord. I guess she truly wants her side of the story to keep 🙃 You can't change history, can you?

BlinkDL (@blinkdl_ai) 's Twitter Profile Photo

p.s. I think arXiv papers can be the next source of reasoning data: (1) Locate difficult yet predictable tokens (2) Use them for RL (3) "Solving" papers will be more than enough to solve the badly-named "Humanity's Last Exam"🙂