Update README.md

README.md

@@ -49,10 +49,6 @@ More RWKV projects: https://github.com/search?o=desc&q=rwkv&s=updated&type=Repos
 
 A cool paper (Spiking Neural Network) using RWKV: https://github.com/ridgerchu/SpikeGPT
 
-ChatRWKV with RWKV 14B ctx8192:
-
-![RWKV-chat](RWKV-chat.png)
-
 You are welcome to join the RWKV discord https://discord.gg/bDSBUMeFpc to build upon it. We have plenty of potential compute (A100 40Gs) now (thanks to Stability and EleutherAI), so if you have interesting ideas I can run them.
 
 ![RWKV-eval2](RWKV-eval2.png)
@@ -61,6 +57,10 @@ RWKV [loss vs token position] for 10000 ctx4k+ documents in Pile. RWKV 1B5-4k is
 
 ![RWKV-ctxlen](RWKV-ctxlen.png)
 
+ChatRWKV with RWKV 14B ctx8192:
+
+![RWKV-chat](RWKV-chat.png)
+
 I believe RNN is a better candidate for fundamental models, because: (1) It's more friendly for ASICs (no kv cache). (2) It's more friendly for RL. (3) When we write, our brain is more similar to RNN. (4) The universe is like an RNN too (because of locality). Transformers are non-local models.
 
 RWKV-3 1.5B on A40 (tf32) = always 0.015 sec/token, tested using simple pytorch code (no CUDA), GPU utilization 45%, VRAM 7823M