File size: 8,117 Bytes
0011066
 
c504c59
 
69b5233
 
0011066
c504c59
13edca7
c504c59
13edca7
038ebdd
c504c59
13edca7
 
 
d26f379
 
c504c59
 
 
 
 
6ebc83d
c504c59
 
 
6d707b7
 
c504c59
 
8fd22c7
6d707b7
c504c59
6d707b7
 
c504c59
6d707b7
 
 
 
 
 
c504c59
13edca7
 
c504c59
 
6d707b7
 
 
 
 
 
 
 
c504c59
6d707b7
 
 
c504c59
 
 
 
8fd22c7
c504c59
447b46e
 
 
 
 
 
 
 
 
c504c59
6d707b7
 
c504c59
6d707b7
 
 
 
 
 
c504c59
13edca7
 
6d707b7
c504c59
6d707b7
 
 
 
 
 
 
 
c504c59
6d707b7
 
 
c504c59
 
 
 
6d707b7
 
c504c59
6d707b7
 
 
 
 
 
c504c59
13edca7
 
c504c59
6d707b7
 
 
 
 
 
 
 
c504c59
6d707b7
 
 
c504c59
 
6d707b7
c504c59
 
 
 
 
 
 
 
 
 
 
 
 
13edca7
c504c59
 
 
 
 
13edca7
1b34338
c504c59
 
 
 
 
 
 
 
 
 
 
 
 
 
 
13edca7
c504c59
 
 
 
 
 
 
 
 
 
 
870f6da
 
c076b6d
870f6da
13edca7
870f6da
c504c59
73dbd14
 
 
 
c504c59
 
 
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
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
---
license: apache-2.0
language:
- en
datasets:
- togethercomputer/RedPajama-Data-1T
---

# RedPajama-INCITE-7B-Base

RedPajama-INCITE-7B-Base was developed by Together and leaders from the open-source AI community including Ontocord.ai, ETH DS3Lab, AAI CERC, Université de Montréal, MILA - Québec AI Institute, Stanford Center for Research on Foundation Models (CRFM), Stanford Hazy Research research group and LAION. 
The training was done on 3,072 V100 GPUs provided as part of the INCITE 2023 project on Scalable Foundation Models for Transferrable Generalist AI, awarded to MILA, LAION, and EleutherAI in fall 2022, with support from the Oak Ridge Leadership Computing Facility (OLCF) and INCITE program. 

  - Base Model: [RedPajama-INCITE-7B-Base](https://huggingface.co/togethercomputer/RedPajama-INCITE-7B-Base)
  - Instruction-tuned Version: [RedPajama-INCITE-7B-Instruct](https://huggingface.co/togethercomputer/RedPajama-INCITE-7B-Instruct)
  - Chat Version: [RedPajama-INCITE-7B-Chat](https://huggingface.co/togethercomputer/RedPajama-INCITE-7B-Chat)


## Model Details
- **Developed by**: Together Computer.
- **Model type**: Language Model
- **Language(s)**: English
- **License**: Apache 2.0
- **Model Description**: A 6.9B parameter pretrained language model.

# Quick Start

Please note that the model requires `transformers` version >= 4.25.1.

## GPU Inference

This requires a GPU with 16GB memory.

```python
import torch
import transformers
from transformers import AutoTokenizer, AutoModelForCausalLM

MIN_TRANSFORMERS_VERSION = '4.25.1'

# check transformers version
assert transformers.__version__ >= MIN_TRANSFORMERS_VERSION, f'Please upgrade transformers to version {MIN_TRANSFORMERS_VERSION} or higher.'

# init
tokenizer = AutoTokenizer.from_pretrained("togethercomputer/RedPajama-INCITE-7B-Base")
model = AutoModelForCausalLM.from_pretrained("togethercomputer/RedPajama-INCITE-7B-Base", torch_dtype=torch.float16)
model = model.to('cuda:0')
# infer
prompt = "Alan Turing is"
inputs = tokenizer(prompt, return_tensors='pt').to(model.device)
input_length = inputs.input_ids.shape[1]
outputs = model.generate(
    **inputs, max_new_tokens=128, do_sample=True, temperature=0.7, top_p=0.7, top_k=50, return_dict_in_generate=True
)
token = outputs.sequences[0, input_length:]
output_str = tokenizer.decode(token)
print(output_str)
"""
widely considered to be the father of modern computer science and artificial intelligence. He was a brilliant mathematician and cryptographer, who worked for the British government during World War II. He was instrumental in breaking the German Enigma code, and is credited with helping to shorten the war by two years...
"""
```

## GPU Inference in Int8

This requires a GPU with 12GB memory.

To run inference with int8, please ensure you have installed accelerate and bitandbytes. You can install them with the following command:

```bash
pip install accelerate
pip install bitsandbytes
```

Then you can run inference with int8 as follows:

```python
import torch
import transformers
from transformers import AutoTokenizer, AutoModelForCausalLM

MIN_TRANSFORMERS_VERSION = '4.25.1'

# check transformers version
assert transformers.__version__ >= MIN_TRANSFORMERS_VERSION, f'Please upgrade transformers to version {MIN_TRANSFORMERS_VERSION} or higher.'

# init
tokenizer = AutoTokenizer.from_pretrained("togethercomputer/RedPajama-INCITE-7B-Base")
model = AutoModelForCausalLM.from_pretrained("togethercomputer/RedPajama-INCITE-7B-Base", device_map='auto', torch_dtype=torch.float16, load_in_8bit=True)

# infer
prompt = "Alan Turing is"
inputs = tokenizer(prompt, return_tensors='pt').to(model.device)
input_length = inputs.input_ids.shape[1]
outputs = model.generate(
    **inputs, max_new_tokens=128, do_sample=True, temperature=0.7, top_p=0.7, top_k=50, return_dict_in_generate=True
)
token = outputs.sequences[0, input_length:]
output_str = tokenizer.decode(token)
print(output_str)
"""
a very well-known name in the world of computer science. It is named after the mathematician Alan Turing. He is famous for his work on the Enigma machine, which was used by the Germans during World War II....
"""```

## CPU Inference

```python
import torch
import transformers
from transformers import AutoTokenizer, AutoModelForCausalLM

MIN_TRANSFORMERS_VERSION = '4.25.1'

# check transformers version
assert transformers.__version__ >= MIN_TRANSFORMERS_VERSION, f'Please upgrade transformers to version {MIN_TRANSFORMERS_VERSION} or higher.'

# init
tokenizer = AutoTokenizer.from_pretrained("togethercomputer/RedPajama-INCITE-7B-Base")
model = AutoModelForCausalLM.from_pretrained("togethercomputer/RedPajama-INCITE-7B-Base", torch_dtype=torch.bfloat16)
# infer
prompt = "Alan Turing is"
inputs = tokenizer(prompt, return_tensors='pt').to(model.device)
input_length = inputs.input_ids.shape[1]
outputs = model.generate(
    **inputs, max_new_tokens=128, do_sample=True, temperature=0.7, top_p=0.7, top_k=50, return_dict_in_generate=True
)
token = outputs.sequences[0, input_length:]
output_str = tokenizer.decode(token)
print(output_str)
"""
one of the most important figures in the history of computing. He is best known for his work on the development of the modern computer and for his code-breaking work during World War II. He was also a brilliant mathematician and philosopher.
"""
```

Please note that since `LayerNormKernelImpl` is not implemented in fp16 for CPU, we use `bfloat16` for CPU inference.

# Uses

## Direct Use 

Excluded uses are described below.

### Misuse, Malicious Use, and Out-of-Scope Use

It is the responsibility of the end user to ensure that the model is used in a responsible and ethical manner.

#### Out-of-Scope Use

`RedPajama-INCITE-7B-Base` is a language model and may not perform well for other use cases outside of its intended scope. 
For example, it may not be suitable for use in safety-critical applications or for making decisions that have a significant impact on individuals or society. 
It is important to consider the limitations of the model and to only use it for its intended purpose.

#### Misuse and Malicious Use

`RedPajama-INCITE-7B-Base` is designed for language modeling.
Misuse of the model, such as using it to engage in illegal or unethical activities, is strictly prohibited and goes against the principles of the project.

Using the model to generate content that is cruel to individuals is a misuse of this model. This includes, but is not limited to:

- Generating fake news, misinformation, or propaganda
- Promoting hate speech, discrimination, or violence against individuals or groups
- Impersonating individuals or organizations without their consent
- Engaging in cyberbullying or harassment
- Defamatory content
- Spamming or scamming
- Sharing confidential or sensitive information without proper authorization
- Violating the terms of use of the model or the data used to train it
- Creating automated bots for malicious purposes such as spreading malware, phishing scams, or spamming

## Limitations

`RedPajama-INCITE-7B-Base`, like other language models, has limitations that should be taken into consideration. 
For example, the model may not always provide accurate or relevant answers, particularly for questions that are complex, ambiguous, or outside of its training data. 
We therefore welcome contributions from individuals and organizations, and encourage collaboration towards creating a more robust and inclusive chatbot.

## Training

**Training Data**

Please refer to [togethercomputer/RedPajama-Data-1T](https://huggingface.co/datasets/togethercomputer/RedPajama-Data-1T)

**Training Procedure**

- **Hardware:** 512 nodes of 6xV100 (IBM Power9), on the OLCF Summit cluster
- **Optimizer:** Apex FusedAdam
- **Parallelism:** Pipeline parallel 12, tensor parallel 2
- **Gradient Accumulations**: 8 (global batch size 4M tokens)
- **Num of Tokens:** 1T Tokens
- **Learning rate:** 0.00012

## Benchmark

Please refer to our [blog post](https://together.xyz) for benchmark results.

## Community

Join us on [Together Discord](https://discord.gg/6ZVDU8tTD4)