Update README.md

README.md

@@ -47,16 +47,16 @@ Efficient inference project: https://github.com/BlinkDL/Albatross
 
 RWKV APP: https://github.com/RWKV-APP/RWKV_APP (local inference on Android/iOS)
 
-Please use **latest G#a3 models** if available (better at everything).
+Please always use **latest G#a# models** (better at everything).
 
 ```
 Gxx = Data Version
 
 G0x = less than 1 epoch, as training 1 epoch for a large model is expensive :(
-G0 G0a G0b ... = adding more (newer and better) data, so G0a has better quality (but less) data than G1
+G0 G0a G0a2 G0a3 ... G0b ... = adding more (newer and better) data, so G0a has better quality (but less) data than G1
 
 G1x = more than 1 epoch
-G1 G1a G1b ... = adding more (newer and better) data, note G1a has better quality (and more) data than G0a
+G1 G1a G1a2 G1a3 ... G1b ... = adding more (newer and better) data, note G1a has better quality (and more) data than G0a
 ```
 
 Decoding (note: this is for RWKV pip pkg, which apply temp after topp):