Update README.md

main
PENG Bo 3 years ago committed by GitHub
parent e0dc08a2ce
commit 9917078f93
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

@ -183,7 +183,15 @@ ss = json.dumps({"meta": meta, "text": text}, ensure_ascii=False)
out.write(ss + "\n")
```
## New ideas (just to record some new ideas)
## Towards RWKV-5 (just to record some new ideas)
### List of some ideas
1. Use complex number (so, rotation instead of decay) in some channels.
2. Inject some trainable and interpolable positional encoding?
### Misc
I have an idea to improve tokenization. We can hardcode some channels to have meanings. Example:
@ -213,7 +221,7 @@ Better: define emb_space emb_capitalize_first emb_capitalize_all to be a functio
Maybe the Best: let 'abc' ' abc' etc. to share the last 90% of their embeddings.
At this moment, all our tokenizers spend too many items to represent all variations of 'abc' ' abc' ' Abc' etc. Moreover the model cannot discover that these are actually similar if some of these variations are rare in the dataset. My method can solve this. I plan to test this in a new version of RWKV.
At this moment, all our tokenizers spend too many items to represent all variations of 'abc' ' abc' ' Abc' etc. Moreover the model cannot discover that these are actually similar if some of these variations are rare in the dataset. The method here can improve this. I plan to test this in a new version of RWKV.
## How it works

Loading…
Cancel
Save