File size: 758 Bytes
c9e7121 701052d c9e7121 701052d c9e7121 701052d c7208f5 701052d 8f1b947 82ffb69 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 |
---
language:
- en
tags:
- pytorch
- text-generation
- causal-lm
- rwkv
license: apache-2.0
datasets:
- the_pile
---
# RWKV-4 14B
## Model Description
RWKV-4 14B is a L40-D5120 causal language model trained on the Pile. See https://github.com/BlinkDL/RWKV-LM for details.
** Note: It's a BF16 model, and it may overflow if you are using FP16 (fixable by rescaling the weights). **
At this moment you have to use my Github code (https://github.com/BlinkDL/RWKV-LM) to run it.
ctx_len = 1024
n_layer = 40
n_embd = 5120
Preview checkpoint: RWKV-4-Pile-14B-20221231-4514.pth : Trained on the Pile for 186B tokens.
* Pile loss 1.8014
* LAMBADA ppl 3.92, acc 69.96%
* PIQA acc 77.42%
* SC2016 acc 74.18%
* Hellaswag acc_norm 68.47%
* WinoGrande acc 62.35%
|