BlinkDL commited on
Commit
e102027
1 Parent(s): 8304362

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +6 -13
README.md CHANGED
@@ -26,16 +26,9 @@ ctx_len = 1024
26
  n_layer = 32
27
  n_embd = 2560
28
 
29
- Preview checkpoint: RWKV-4-Pile-3B-20220921-3047.pth : Trained on the Pile for 125B tokens.
30
- * Pile loss 2.0026
31
- * LAMBADA ppl 5.72, acc 61.36%
32
- * PIQA acc 73.39%
33
- * SC2016 acc 68.84%
34
- * Hellaswag acc_norm 56.57%
35
-
36
- Preview checkpoint: RWKV-4-Pile-3B-20220915-1207.pth : Trained on the Pile for 50B tokens.
37
- * Pile loss 2.0902
38
- * LAMBADA ppl 7.01, acc 57.11%
39
- * PIQA acc 72.52%
40
- * SC2016 acc 68.36%
41
- * Hellaswag acc_norm 52.17%
 
26
  n_layer = 32
27
  n_embd = 2560
28
 
29
+ Preview checkpoint: RWKV-4-Pile-3B-20220923-3822.pth : Trained on the Pile for 157B tokens.
30
+ * Pile loss 1.9857
31
+ * LAMBADA ppl 5.61, acc 61.87%
32
+ * PIQA acc 73.78%
33
+ * SC2016 acc 69.80%
34
+ * Hellaswag acc_norm 57.46%