Update README.md
Browse files
README.md
CHANGED
|
@@ -24,7 +24,20 @@ args.n_embd = 5120
|
|
| 24 |
Use https://github.com/BlinkDL/ChatRWKV to run it.
|
| 25 |
|
| 26 |
RWKV-4-Pile-14B-2023xxxx-ctx8192-testxxx.pth : Fine-tuned to ctx_len 8192.
|
| 27 |
-
* The best model.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 28 |
|
| 29 |
RWKV-4-Pile-14B-20230213-8019.pth : Trained on the Pile for 331B tokens
|
| 30 |
* Pile loss 1.7579 (ctx_len 1024)
|
|
|
|
| 24 |
Use https://github.com/BlinkDL/ChatRWKV to run it.
|
| 25 |
|
| 26 |
RWKV-4-Pile-14B-2023xxxx-ctx8192-testxxx.pth : Fine-tuned to ctx_len 8192.
|
| 27 |
+
* The best general model.
|
| 28 |
+
|
| 29 |
+
################################
|
| 30 |
+
|
| 31 |
+
RWKV 14B Alpaca test model (both are finetuned from ctx8192. i think the ctx1024 version should be better for usual Q&A. please test & compare)
|
| 32 |
+
|
| 33 |
+
https://huggingface.co/BlinkDL/rwkv-4-pile-14b/blob/main/RWKV-4-Pile-14B-Instruct-test4-20230327-ctx4096.pth
|
| 34 |
+
|
| 35 |
+
(Update ChatRWKV v2 to latest version first) It's recommended to use +i for "Alpaca Instruct". Examples:
|
| 36 |
+
```
|
| 37 |
+
+i Explain the following metaphor: "Life is like cats".
|
| 38 |
+
+i write a python function to read data from an excel file.
|
| 39 |
+
```
|
| 40 |
+
################################
|
| 41 |
|
| 42 |
RWKV-4-Pile-14B-20230213-8019.pth : Trained on the Pile for 331B tokens
|
| 43 |
* Pile loss 1.7579 (ctx_len 1024)
|