File size: 979 Bytes
c9e7121
701052d
 
 
 
 
 
 
c9e7121
701052d
 
 
c9e7121
701052d
 
 
a05e790
 
 
 
701052d
 
 
 
50b472b
701052d
1a3be9c
807a65e
1a3be9c
a91be67
 
a3ce566
 
 
 
 
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
37
---
language:
- en
tags:
- pytorch
- text-generation
- causal-lm
- rwkv
license: apache-2.0
datasets:
- the_pile

---

# RWKV-4 14B

# WARNING: 4 / 4a / 4b models ARE NOT compatible. Use RWKV-4 models unless you know what you are doing.
# WARNING: 4 / 4a / 4b models ARE NOT compatible. Use RWKV-4 models unless you know what you are doing.
# WARNING: 4 / 4a / 4b models ARE NOT compatible. Use RWKV-4 models unless you know what you are doing.

## 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.

Use https://github.com/BlinkDL/ChatRWKV to run it.

RWKV-4-Pile-14B-2023xxxx-ctx4096-testxxx.pth : Fine-tuned to ctx_len 4096.
* Highly recommended. It's great.

RWKV-4-Pile-14B-20230213-8019.pth : Trained on the Pile for 331B tokens
* Pile loss 1.7579 (ctx_len 1024)
* LAMBADA ppl 3.81, acc 71.05%
* PIQA acc 77.42%
* SC2016 acc 75.57%
* Hellaswag acc_norm 70.24%
* WinoGrande acc 62.98%