Text Generation
PyTorch
12 languages
causal-lm
rwkv
File size: 2,447 Bytes
b6a5c87
5d58c65
 
 
 
 
ef91075
5d58c65
 
 
 
 
 
 
 
 
 
 
 
b6a5c87
5d58c65
8411e7f
ef91075
e224013
 
b6a5c87
5d58c65
88a9769
d9a21d0
d8368ac
f34b018
88a9769
5d58c65
5f12317
 
8411e7f
 
9b6d58f
 
11fe6c3
 
 
 
8411e7f
81a1a9b
361e909
 
5d58c65
 
 
 
4696f7d
ce7047c
4720116
050fd2b
 
 
 
 
1a6472b
 
54a2831
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1a6472b
 
 
 
 
 
 
 
 
54a2831
1a6472b
 
 
 
 
 
 
 
 
 
 
 
bdeecd1
 
 
 
 
 
 
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
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
---
language:
- en
- zh
- fr
- es
- de
- pt
- ru
- it
- ja
- ko
- vi
- ar
tags:
- pytorch
- text-generation
- causal-lm
- rwkv
license: apache-2.0
datasets:
- cerebras/SlimPajama-627B
- EleutherAI/pile
- bigcode/starcoderdata
- oscar-corpus/OSCAR-2301
---

# RWKV-5 World

Use rwkv pip package 0.8.22+ for RWKV-5 inference: https://pypi.org/project/rwkv/ (pipeline = PIPELINE(model, "rwkv_vocab_v20230424") for rwkv-world models)

Online 7B Demo: https://huggingface.co/spaces/BlinkDL/RWKV-Gradio-2

Online 1.5B Demo: https://huggingface.co/spaces/BlinkDL/RWKV-Gradio-1

GUI: https://github.com/josStorer/RWKV-Runner (see Releases)

Convert to HF formet: https://github.com/BBuf/RWKV-World-HF-Tokenizer

For developer: https://github.com/BlinkDL/ChatRWKV/blob/main/API_DEMO_CHAT.py

https://github.com/BlinkDL/ChatRWKV/blob/main/RWKV_v5_demo.py

How it works: https://twitter.com/BlinkDL_AI/status/1685230712247795713

https://www.rwkv.com/

## Model Description

RWKV-5 trained on 100+ world languages (70% English, 15% multilang, 15% code).

World = Some_Pile + Some_SlimPajama + Some_StarCoder + Some_OSCAR + All_Wikipedia + All_ChatGPT_Data_I_can_find

RWKV-5 training: set --my_testing "r2r4" in latest RWKV-LM v4neo: https://github.com/BlinkDL/RWKV-LM

World v1 = 0.59T tokens

World v2 = 1.12T tokens

Imagine what happens when we use more data :)

Recommended fine-tuning format (use \n for newlines):
```
User: xxxxxxxxxxxxxxx

Assistant: xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxx

User: xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxx

Assistant: xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxx
```

A good chat prompt (better replace \n\n in xxx to \n, such that there will be no newlines in xxx):
```
User: hi

Assistant: Hi. I am your assistant and I will provide expert full response in full details. Please feel free to ask any question and I will always answer it.

User: xxx

Assistant:
```
QA prompt (better replace \n\n in xxx to \n, such that there will be no newlines in xxx):
```
Question: xxx

Answer:
```
and
```
Instruction: xxx

Input: xxx

Response:
```

!!! There should not be any space after your final ":" or you will upset the tokenizer and see non-English reponse !!!

!!! There should not be any space after your final ":" or you will upset the tokenizer and see non-English reponse !!!

!!! There should not be any space after your final ":" or you will upset the tokenizer and see non-English reponse !!!