RiversHaveWings commited on
Commit
856c049
1 Parent(s): e259381

Add LICENSE, README.md, and USE_POLICY.md from original repo

Browse files
Files changed (3) hide show
  1. LICENSE +114 -0
  2. README.md +1208 -0
  3. USE_POLICY.md +51 -0
LICENSE ADDED
@@ -0,0 +1,114 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ LLAMA 3.1 COMMUNITY LICENSE AGREEMENT
2
+ Llama 3.1 Version Release Date: July 23, 2024
3
+
4
+ “Agreement” means the terms and conditions for use, reproduction, distribution and modification of the
5
+ Llama Materials set forth herein.
6
+
7
+ “Documentation” means the specifications, manuals and documentation accompanying Llama 3.1
8
+ distributed by Meta at https://llama.meta.com/doc/overview.
9
+
10
+ “Licensee” or “you” means you, or your employer or any other person or entity (if you are entering into
11
+ this Agreement on such person or entity’s behalf), of the age required under applicable laws, rules or
12
+ regulations to provide legal consent and that has legal authority to bind your employer or such other
13
+ person or entity if you are entering in this Agreement on their behalf.
14
+
15
+ “Llama 3.1” means the foundational large language models and software and algorithms, including
16
+ machine-learning model code, trained model weights, inference-enabling code, training-enabling code,
17
+ fine-tuning enabling code and other elements of the foregoing distributed by Meta at
18
+ https://llama.meta.com/llama-downloads.
19
+
20
+ “Llama Materials” means, collectively, Meta’s proprietary Llama 3.1 and Documentation (and any
21
+ portion thereof) made available under this Agreement.
22
+
23
+ “Meta” or “we” means Meta Platforms Ireland Limited (if you are located in or, if you are an entity, your
24
+ principal place of business is in the EEA or Switzerland) and Meta Platforms, Inc. (if you are located
25
+ outside of the EEA or Switzerland).
26
+
27
+ By clicking “I Accept” below or by using or distributing any portion or element of the Llama Materials,
28
+ you agree to be bound by this Agreement.
29
+
30
+ 1. License Rights and Redistribution.
31
+
32
+ a. Grant of Rights. You are granted a non-exclusive, worldwide, non-transferable and royalty-free
33
+ limited license under Meta’s intellectual property or other rights owned by Meta embodied in the Llama
34
+ Materials to use, reproduce, distribute, copy, create derivative works of, and make modifications to the
35
+ Llama Materials.
36
+
37
+ b. Redistribution and Use.
38
+
39
+ i. If you distribute or make available the Llama Materials (or any derivative works
40
+ thereof), or a product or service (including another AI model) that contains any of them, you shall (A)
41
+ provide a copy of this Agreement with any such Llama Materials; and (B) prominently display “Built with
42
+ Llama” on a related website, user interface, blogpost, about page, or product documentation. If you use
43
+ the Llama Materials or any outputs or results of the Llama Materials to create, train, fine tune, or
44
+ otherwise improve an AI model, which is distributed or made available, you shall also include “Llama” at
45
+ the beginning of any such AI model name.
46
+
47
+ ii. If you receive Llama Materials, or any derivative works thereof, from a Licensee as part
48
+ of an integrated end user product, then Section 2 of this Agreement will not apply to you.
49
+
50
+ iii. You must retain in all copies of the Llama Materials that you distribute the following
51
+ attribution notice within a “Notice” text file distributed as a part of such copies: “Llama 3.1 is
52
+ licensed under the Llama 3.1 Community License, Copyright © Meta Platforms, Inc. All Rights
53
+ Reserved.”
54
+
55
+ iv. Your use of the Llama Materials must comply with applicable laws and regulations
56
+ (including trade compliance laws and regulations) and adhere to the Acceptable Use Policy for the Llama
57
+ Materials (available at https://llama.meta.com/llama3_1/use-policy), which is hereby incorporated by
58
+ reference into this Agreement.
59
+
60
+ 2. Additional Commercial Terms. If, on the Llama 3.1 version release date, the monthly active users
61
+ of the products or services made available by or for Licensee, or Licensee’s affiliates, is greater than 700
62
+ million monthly active users in the preceding calendar month, you must request a license from Meta,
63
+ which Meta may grant to you in its sole discretion, and you are not authorized to exercise any of the
64
+ rights under this Agreement unless or until Meta otherwise expressly grants you such rights.
65
+
66
+ 3. Disclaimer of Warranty. UNLESS REQUIRED BY APPLICABLE LAW, THE LLAMA MATERIALS AND ANY
67
+ OUTPUT AND RESULTS THEREFROM ARE PROVIDED ON AN “AS IS” BASIS, WITHOUT WARRANTIES OF
68
+ ANY KIND, AND META DISCLAIMS ALL WARRANTIES OF ANY KIND, BOTH EXPRESS AND IMPLIED,
69
+ INCLUDING, WITHOUT LIMITATION, ANY WARRANTIES OF TITLE, NON-INFRINGEMENT,
70
+ MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE. YOU ARE SOLELY RESPONSIBLE FOR
71
+ DETERMINING THE APPROPRIATENESS OF USING OR REDISTRIBUTING THE LLAMA MATERIALS AND
72
+ ASSUME ANY RISKS ASSOCIATED WITH YOUR USE OF THE LLAMA MATERIALS AND ANY OUTPUT AND
73
+ RESULTS.
74
+
75
+ 4. Limitation of Liability. IN NO EVENT WILL META OR ITS AFFILIATES BE LIABLE UNDER ANY THEORY OF
76
+ LIABILITY, WHETHER IN CONTRACT, TORT, NEGLIGENCE, PRODUCTS LIABILITY, OR OTHERWISE, ARISING
77
+ OUT OF THIS AGREEMENT, FOR ANY LOST PROFITS OR ANY INDIRECT, SPECIAL, CONSEQUENTIAL,
78
+ INCIDENTAL, EXEMPLARY OR PUNITIVE DAMAGES, EVEN IF META OR ITS AFFILIATES HAVE BEEN ADVISED
79
+ OF THE POSSIBILITY OF ANY OF THE FOREGOING.
80
+
81
+ 5. Intellectual Property.
82
+
83
+ a. No trademark licenses are granted under this Agreement, and in connection with the Llama
84
+ Materials, neither Meta nor Licensee may use any name or mark owned by or associated with the other
85
+ or any of its affiliates, except as required for reasonable and customary use in describing and
86
+ redistributing the Llama Materials or as set forth in this Section 5(a). Meta hereby grants you a license to
87
+ use “Llama” (the “Mark”) solely as required to comply with the last sentence of Section 1.b.i. You will
88
+ comply with Meta’s brand guidelines (currently accessible at
89
+ https://about.meta.com/brand/resources/meta/company-brand/ ). All goodwill arising out of your use
90
+ of the Mark will inure to the benefit of Meta.
91
+
92
+ b. Subject to Meta’s ownership of Llama Materials and derivatives made by or for Meta, with
93
+ respect to any derivative works and modifications of the Llama Materials that are made by you, as
94
+ between you and Meta, you are and will be the owner of such derivative works and modifications.
95
+
96
+ c. If you institute litigation or other proceedings against Meta or any entity (including a
97
+ cross-claim or counterclaim in a lawsuit) alleging that the Llama Materials or Llama 3.1 outputs or
98
+ results, or any portion of any of the foregoing, constitutes infringement of intellectual property or other
99
+ rights owned or licensable by you, then any licenses granted to you under this Agreement shall
100
+ terminate as of the date such litigation or claim is filed or instituted. You will indemnify and hold
101
+ harmless Meta from and against any claim by any third party arising out of or related to your use or
102
+ distribution of the Llama Materials.
103
+
104
+ 6. Term and Termination. The term of this Agreement will commence upon your acceptance of this
105
+ Agreement or access to the Llama Materials and will continue in full force and effect until terminated in
106
+ accordance with the terms and conditions herein. Meta may terminate this Agreement if you are in
107
+ breach of any term or condition of this Agreement. Upon termination of this Agreement, you shall delete
108
+ and cease use of the Llama Materials. Sections 3, 4 and 7 shall survive the termination of this
109
+ Agreement.
110
+
111
+ 7. Governing Law and Jurisdiction. This Agreement will be governed and construed under the laws of
112
+ the State of California without regard to choice of law principles, and the UN Convention on Contracts
113
+ for the International Sale of Goods does not apply to this Agreement. The courts of California shall have
114
+ exclusive jurisdiction of any dispute arising out of this Agreement.
README.md ADDED
@@ -0,0 +1,1208 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - en
4
+ - de
5
+ - fr
6
+ - it
7
+ - pt
8
+ - hi
9
+ - es
10
+ - th
11
+ pipeline_tag: text-generation
12
+ tags:
13
+ - facebook
14
+ - meta
15
+ - pytorch
16
+ - llama
17
+ - llama-3
18
+ license: llama3.1
19
+ extra_gated_prompt: >-
20
+ ### LLAMA 3.1 COMMUNITY LICENSE AGREEMENT
21
+
22
+ Llama 3.1 Version Release Date: July 23, 2024
23
+
24
+ "Agreement" means the terms and conditions for use, reproduction, distribution
25
+ and modification of the Llama Materials set forth herein.
26
+
27
+ "Documentation" means the specifications, manuals and documentation
28
+ accompanying Llama 3.1 distributed by Meta at
29
+ https://llama.meta.com/doc/overview.
30
+
31
+ "Licensee" or "you" means you, or your employer or any other person or entity
32
+ (if you are entering into this Agreement on such person or entity’s behalf),
33
+ of the age required under applicable laws, rules or regulations to provide
34
+ legal consent and that has legal authority to bind your employer or such other
35
+ person or entity if you are entering in this Agreement on their behalf.
36
+
37
+ "Llama 3.1" means the foundational large language models and software and
38
+ algorithms, including machine-learning model code, trained model weights,
39
+ inference-enabling code, training-enabling code, fine-tuning enabling code and
40
+ other elements of the foregoing distributed by Meta at
41
+ https://llama.meta.com/llama-downloads.
42
+
43
+ "Llama Materials" means, collectively, Meta’s proprietary Llama 3.1 and
44
+ Documentation (and any portion thereof) made available under this Agreement.
45
+
46
+ "Meta" or "we" means Meta Platforms Ireland Limited (if you are located in or,
47
+ if you are an entity, your principal place of business is in the EEA or
48
+ Switzerland) and Meta Platforms, Inc. (if you are located outside of the EEA
49
+ or Switzerland).
50
+
51
+ 1. License Rights and Redistribution.
52
+
53
+ a. Grant of Rights. You are granted a non-exclusive, worldwide,
54
+ non-transferable and royalty-free limited license under Meta’s intellectual
55
+ property or other rights owned by Meta embodied in the Llama Materials to use,
56
+ reproduce, distribute, copy, create derivative works of, and make
57
+ modifications to the Llama Materials.
58
+
59
+ b. Redistribution and Use.
60
+
61
+ i. If you distribute or make available the Llama Materials (or any derivative
62
+ works thereof), or a product or service (including another AI model) that
63
+ contains any of them, you shall (A) provide a copy of this Agreement with any
64
+ such Llama Materials; and (B) prominently display “Built with Llama” on a
65
+ related website, user interface, blogpost, about page, or product
66
+ documentation. If you use the Llama Materials or any outputs or results of the
67
+ Llama Materials to create, train, fine tune, or otherwise improve an AI model,
68
+ which is distributed or made available, you shall also include “Llama” at the
69
+ beginning of any such AI model name.
70
+
71
+ ii. If you receive Llama Materials, or any derivative works thereof, from a
72
+ Licensee as part of an integrated end user product, then Section 2 of this
73
+ Agreement will not apply to you.
74
+
75
+ iii. You must retain in all copies of the Llama Materials that you distribute
76
+ the following attribution notice within a “Notice” text file distributed as a
77
+ part of such copies: “Llama 3.1 is licensed under the Llama 3.1 Community
78
+ License, Copyright © Meta Platforms, Inc. All Rights Reserved.”
79
+
80
+ iv. Your use of the Llama Materials must comply with applicable laws and
81
+ regulations (including trade compliance laws and regulations) and adhere to
82
+ the Acceptable Use Policy for the Llama Materials (available at
83
+ https://llama.meta.com/llama3_1/use-policy), which is hereby incorporated by
84
+ reference into this Agreement.
85
+
86
+ 2. Additional Commercial Terms. If, on the Llama 3.1 version release date, the
87
+ monthly active users of the products or services made available by or for
88
+ Licensee, or Licensee’s affiliates, is greater than 700 million monthly active
89
+ users in the preceding calendar month, you must request a license from Meta,
90
+ which Meta may grant to you in its sole discretion, and you are not authorized
91
+ to exercise any of the rights under this Agreement unless or until Meta
92
+ otherwise expressly grants you such rights.
93
+
94
+ 3. Disclaimer of Warranty. UNLESS REQUIRED BY APPLICABLE LAW, THE LLAMA
95
+ MATERIALS AND ANY OUTPUT AND RESULTS THEREFROM ARE PROVIDED ON AN “AS IS”
96
+ BASIS, WITHOUT WARRANTIES OF ANY KIND, AND META DISCLAIMS ALL WARRANTIES OF
97
+ ANY KIND, BOTH EXPRESS AND IMPLIED, INCLUDING, WITHOUT LIMITATION, ANY
98
+ WARRANTIES OF TITLE, NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A
99
+ PARTICULAR PURPOSE. YOU ARE SOLELY RESPONSIBLE FOR DETERMINING THE
100
+ APPROPRIATENESS OF USING OR REDISTRIBUTING THE LLAMA MATERIALS AND ASSUME ANY
101
+ RISKS ASSOCIATED WITH YOUR USE OF THE LLAMA MATERIALS AND ANY OUTPUT AND
102
+ RESULTS.
103
+
104
+ 4. Limitation of Liability. IN NO EVENT WILL META OR ITS AFFILIATES BE LIABLE
105
+ UNDER ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, TORT, NEGLIGENCE, PRODUCTS
106
+ LIABILITY, OR OTHERWISE, ARISING OUT OF THIS AGREEMENT, FOR ANY LOST PROFITS
107
+ OR ANY INDIRECT, SPECIAL, CONSEQUENTIAL, INCIDENTAL, EXEMPLARY OR PUNITIVE
108
+ DAMAGES, EVEN IF META OR ITS AFFILIATES HAVE BEEN ADVISED OF THE POSSIBILITY
109
+ OF ANY OF THE FOREGOING.
110
+
111
+ 5. Intellectual Property.
112
+
113
+ a. No trademark licenses are granted under this Agreement, and in connection
114
+ with the Llama Materials, neither Meta nor Licensee may use any name or mark
115
+ owned by or associated with the other or any of its affiliates, except as
116
+ required for reasonable and customary use in describing and redistributing the
117
+ Llama Materials or as set forth in this Section 5(a). Meta hereby grants you a
118
+ license to use “Llama” (the “Mark”) solely as required to comply with the last
119
+ sentence of Section 1.b.i. You will comply with Meta’s brand guidelines
120
+ (currently accessible at
121
+ https://about.meta.com/brand/resources/meta/company-brand/ ). All goodwill
122
+ arising out of your use of the Mark will inure to the benefit of Meta.
123
+
124
+ b. Subject to Meta’s ownership of Llama Materials and derivatives made by or
125
+ for Meta, with respect to any derivative works and modifications of the Llama
126
+ Materials that are made by you, as between you and Meta, you are and will be
127
+ the owner of such derivative works and modifications.
128
+
129
+ c. If you institute litigation or other proceedings against Meta or any entity
130
+ (including a cross-claim or counterclaim in a lawsuit) alleging that the Llama
131
+ Materials or Llama 3.1 outputs or results, or any portion of any of the
132
+ foregoing, constitutes infringement of intellectual property or other rights
133
+ owned or licensable by you, then any licenses granted to you under this
134
+ Agreement shall terminate as of the date such litigation or claim is filed or
135
+ instituted. You will indemnify and hold harmless Meta from and against any
136
+ claim by any third party arising out of or related to your use or distribution
137
+ of the Llama Materials.
138
+
139
+ 6. Term and Termination. The term of this Agreement will commence upon your
140
+ acceptance of this Agreement or access to the Llama Materials and will
141
+ continue in full force and effect until terminated in accordance with the
142
+ terms and conditions herein. Meta may terminate this Agreement if you are in
143
+ breach of any term or condition of this Agreement. Upon termination of this
144
+ Agreement, you shall delete and cease use of the Llama Materials. Sections 3,
145
+ 4 and 7 shall survive the termination of this Agreement.
146
+
147
+ 7. Governing Law and Jurisdiction. This Agreement will be governed and
148
+ construed under the laws of the State of California without regard to choice
149
+ of law principles, and the UN Convention on Contracts for the International
150
+ Sale of Goods does not apply to this Agreement. The courts of California shall
151
+ have exclusive jurisdiction of any dispute arising out of this Agreement.
152
+
153
+ ### Llama 3.1 Acceptable Use Policy
154
+
155
+ Meta is committed to promoting safe and fair use of its tools and features,
156
+ including Llama 3.1. If you access or use Llama 3.1, you agree to this
157
+ Acceptable Use Policy (“Policy”). The most recent copy of this policy can be
158
+ found at
159
+ [https://llama.meta.com/llama3_1/use-policy](https://llama.meta.com/llama3_1/use-policy)
160
+
161
+ #### Prohibited Uses
162
+
163
+ We want everyone to use Llama 3.1 safely and responsibly. You agree you will
164
+ not use, or allow others to use, Llama 3.1 to:
165
+ 1. Violate the law or others’ rights, including to:
166
+ 1. Engage in, promote, generate, contribute to, encourage, plan, incite, or further illegal or unlawful activity or content, such as:
167
+ 1. Violence or terrorism
168
+ 2. Exploitation or harm to children, including the solicitation, creation, acquisition, or dissemination of child exploitative content or failure to report Child Sexual Abuse Material
169
+ 3. Human trafficking, exploitation, and sexual violence
170
+ 4. The illegal distribution of information or materials to minors, including obscene materials, or failure to employ legally required age-gating in connection with such information or materials.
171
+ 5. Sexual solicitation
172
+ 6. Any other criminal activity
173
+ 3. Engage in, promote, incite, or facilitate the harassment, abuse, threatening, or bullying of individuals or groups of individuals
174
+ 4. Engage in, promote, incite, or facilitate discrimination or other unlawful or harmful conduct in the provision of employment, employment benefits, credit, housing, other economic benefits, or other essential goods and services
175
+ 5. Engage in the unauthorized or unlicensed practice of any profession including, but not limited to, financial, legal, medical/health, or related professional practices
176
+ 6. Collect, process, disclose, generate, or infer health, demographic, or other sensitive personal or private information about individuals without rights and consents required by applicable laws
177
+ 7. Engage in or facilitate any action or generate any content that infringes, misappropriates, or otherwise violates any third-party rights, including the outputs or results of any products or services using the Llama Materials
178
+ 8. Create, generate, or facilitate the creation of malicious code, malware, computer viruses or do anything else that could disable, overburden, interfere with or impair the proper working, integrity, operation or appearance of a website or computer system
179
+ 2. Engage in, promote, incite, facilitate, or assist in the planning or
180
+ development of activities that present a risk of death or bodily harm to
181
+ individuals, including use of Llama 3.1 related to the following:
182
+ 1. Military, warfare, nuclear industries or applications, espionage, use for materials or activities that are subject to the International Traffic Arms Regulations (ITAR) maintained by the United States Department of State
183
+ 2. Guns and illegal weapons (including weapon development)
184
+ 3. Illegal drugs and regulated/controlled substances
185
+ 4. Operation of critical infrastructure, transportation technologies, or heavy machinery
186
+ 5. Self-harm or harm to others, including suicide, cutting, and eating disorders
187
+ 6. Any content intended to incite or promote violence, abuse, or any infliction of bodily harm to an individual
188
+ 3. Intentionally deceive or mislead others, including use of Llama 3.1 related
189
+ to the following:
190
+ 1. Generating, promoting, or furthering fraud or the creation or promotion of disinformation
191
+ 2. Generating, promoting, or furthering defamatory content, including the creation of defamatory statements, images, or other content
192
+ 3. Generating, promoting, or further distributing spam
193
+ 4. Impersonating another individual without consent, authorization, or legal right
194
+ 5. Representing that the use of Llama 3.1 or outputs are human-generated
195
+ 6. Generating or facilitating false online engagement, including fake reviews and other means of fake online engagement
196
+ 4. Fail to appropriately disclose to end users any known dangers of your AI
197
+ system
198
+
199
+ Please report any violation of this Policy, software “bug,” or other problems
200
+ that could lead to a violation of this Policy through one of the following
201
+ means:
202
+ * Reporting issues with the model: [https://github.com/meta-llama/llama-models/issues](https://github.com/meta-llama/llama-models/issues)
203
+ * Reporting risky content generated by the model:
204
+ developers.facebook.com/llama_output_feedback
205
+ * Reporting bugs and security concerns: facebook.com/whitehat/info
206
+ * Reporting violations of the Acceptable Use Policy or unlicensed uses of Meta Llama 3: [email protected]
207
+ extra_gated_fields:
208
+ First Name: text
209
+ Last Name: text
210
+ Date of birth: date_picker
211
+ Country: country
212
+ Affiliation: text
213
+ Job title:
214
+ type: select
215
+ options:
216
+ - Student
217
+ - Research Graduate
218
+ - AI researcher
219
+ - AI developer/engineer
220
+ - Reporter
221
+ - Other
222
+ geo: ip_location
223
+ By clicking Submit below I accept the terms of the license and acknowledge that the information I provide will be collected stored processed and shared in accordance with the Meta Privacy Policy: checkbox
224
+ extra_gated_description: >-
225
+ The information you provide will be collected, stored, processed and shared in
226
+ accordance with the [Meta Privacy
227
+ Policy](https://www.facebook.com/privacy/policy/).
228
+ extra_gated_button_content: Submit
229
+ library_name: transformers
230
+ ---
231
+
232
+ ## Model Information
233
+
234
+ The Meta Llama 3.1 collection of multilingual large language models (LLMs) is a collection of pretrained and instruction tuned generative models in 8B, 70B and 405B sizes (text in/text out). The Llama 3.1 instruction tuned text only models (8B, 70B, 405B) are optimized for multilingual dialogue use cases and outperform many of the available open source and closed chat models on common industry benchmarks.
235
+
236
+ **Model developer**: Meta
237
+
238
+ **Model Architecture:** Llama 3.1 is an auto-regressive language model that uses an optimized transformer architecture. The tuned versions use supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF) to align with human preferences for helpfulness and safety.
239
+
240
+
241
+ <table>
242
+ <tr>
243
+ <td>
244
+ </td>
245
+ <td><strong>Training Data</strong>
246
+ </td>
247
+ <td><strong>Params</strong>
248
+ </td>
249
+ <td><strong>Input modalities</strong>
250
+ </td>
251
+ <td><strong>Output modalities</strong>
252
+ </td>
253
+ <td><strong>Context length</strong>
254
+ </td>
255
+ <td><strong>GQA</strong>
256
+ </td>
257
+ <td><strong>Token count</strong>
258
+ </td>
259
+ <td><strong>Knowledge cutoff</strong>
260
+ </td>
261
+ </tr>
262
+ <tr>
263
+ <td rowspan="3" >Llama 3.1 (text only)
264
+ </td>
265
+ <td rowspan="3" >A new mix of publicly available online data.
266
+ </td>
267
+ <td>8B
268
+ </td>
269
+ <td>Multilingual Text
270
+ </td>
271
+ <td>Multilingual Text and code
272
+ </td>
273
+ <td>128k
274
+ </td>
275
+ <td>Yes
276
+ </td>
277
+ <td rowspan="3" >15T+
278
+ </td>
279
+ <td rowspan="3" >December 2023
280
+ </td>
281
+ </tr>
282
+ <tr>
283
+ <td>70B
284
+ </td>
285
+ <td>Multilingual Text
286
+ </td>
287
+ <td>Multilingual Text and code
288
+ </td>
289
+ <td>128k
290
+ </td>
291
+ <td>Yes
292
+ </td>
293
+ </tr>
294
+ <tr>
295
+ <td>405B
296
+ </td>
297
+ <td>Multilingual Text
298
+ </td>
299
+ <td>Multilingual Text and code
300
+ </td>
301
+ <td>128k
302
+ </td>
303
+ <td>Yes
304
+ </td>
305
+ </tr>
306
+ </table>
307
+
308
+
309
+ **Supported languages:** English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai.
310
+
311
+ **Llama 3.1 family of models**. Token counts refer to pretraining data only. All model versions use Grouped-Query Attention (GQA) for improved inference scalability.
312
+
313
+ **Model Release Date:** July 23, 2024.
314
+
315
+ **Status:** This is a static model trained on an offline dataset. Future versions of the tuned models will be released as we improve model safety with community feedback.
316
+
317
+ **License:** A custom commercial license, the Llama 3.1 Community License, is available at: [https://github.com/meta-llama/llama-models/blob/main/models/llama3_1/LICENSE](https://github.com/meta-llama/llama-models/blob/main/models/llama3_1/LICENSE)
318
+
319
+ Where to send questions or comments about the model Instructions on how to provide feedback or comments on the model can be found in the model [README](https://github.com/meta-llama/llama3). For more technical information about generation parameters and recipes for how to use Llama 3.1 in applications, please go [here](https://github.com/meta-llama/llama-recipes).
320
+
321
+
322
+ ## Intended Use
323
+
324
+ **Intended Use Cases** Llama 3.1 is intended for commercial and research use in multiple languages. Instruction tuned text only models are intended for assistant-like chat, whereas pretrained models can be adapted for a variety of natural language generation tasks. The Llama 3.1 model collection also supports the ability to leverage the outputs of its models to improve other models including synthetic data generation and distillation. The Llama 3.1 Community License allows for these use cases.
325
+
326
+ **Out-of-scope** Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in any other way that is prohibited by the Acceptable Use Policy and Llama 3.1 Community License. Use in languages beyond those explicitly referenced as supported in this model card**.
327
+
328
+ **<span style="text-decoration:underline;">Note</span>: Llama 3.1 has been trained on a broader collection of languages than the 8 supported languages. Developers may fine-tune Llama 3.1 models for languages beyond the 8 supported languages provided they comply with the Llama 3.1 Community License and the Acceptable Use Policy and in such cases are responsible for ensuring that any uses of Llama 3.1 in additional languages is done in a safe and responsible manner.
329
+
330
+
331
+ ## Hardware and Software
332
+
333
+ **Training Factors** We used custom training libraries, Meta's custom built GPU cluster, and production infrastructure for pretraining. Fine-tuning, annotation, and evaluation were also performed on production infrastructure.
334
+
335
+ **Training utilized a cumulative of** 39.3M GPU hours of computation on H100-80GB (TDP of 700W) type hardware, per the table below. Training time is the total GPU time required for training each model and power consumption is the peak power capacity per GPU device used, adjusted for power usage efficiency.
336
+
337
+
338
+ **Training Greenhouse Gas Emissions** Estimated total location-based greenhouse gas emissions were **11,390** tons CO2eq for training. Since 2020, Meta has maintained net zero greenhouse gas emissions in its global operations and matched 100% of its electricity use with renewable energy, therefore the total market-based greenhouse gas emissions for training were 0 tons CO2eq.
339
+
340
+
341
+ <table>
342
+ <tr>
343
+ <td>
344
+ </td>
345
+ <td><strong>Training Time (GPU hours)</strong>
346
+ </td>
347
+ <td><strong>Training Power Consumption (W)</strong>
348
+ </td>
349
+ <td><strong>Training Location-Based Greenhouse Gas Emissions</strong>
350
+ <p>
351
+ <strong>(tons CO2eq)</strong>
352
+ </td>
353
+ <td><strong>Training Market-Based Greenhouse Gas Emissions</strong>
354
+ <p>
355
+ <strong>(tons CO2eq)</strong>
356
+ </td>
357
+ </tr>
358
+ <tr>
359
+ <td>Llama 3.1 8B
360
+ </td>
361
+ <td>1.46M
362
+ </td>
363
+ <td>700
364
+ </td>
365
+ <td>420
366
+ </td>
367
+ <td>0
368
+ </td>
369
+ </tr>
370
+ <tr>
371
+ <td>Llama 3.1 70B
372
+ </td>
373
+ <td>7.0M
374
+ </td>
375
+ <td>700
376
+ </td>
377
+ <td>2,040
378
+ </td>
379
+ <td>0
380
+ </td>
381
+ </tr>
382
+ <tr>
383
+ <td>Llama 3.1 405B
384
+ </td>
385
+ <td>30.84M
386
+ </td>
387
+ <td>700
388
+ </td>
389
+ <td>8,930
390
+ </td>
391
+ <td>0
392
+ </td>
393
+ </tr>
394
+ <tr>
395
+ <td>Total
396
+ </td>
397
+ <td>39.3M
398
+ <td>
399
+ <ul>
400
+
401
+ </ul>
402
+ </td>
403
+ <td>11,390
404
+ </td>
405
+ <td>0
406
+ </td>
407
+ </tr>
408
+ </table>
409
+
410
+
411
+
412
+ The methodology used to determine training energy use and greenhouse gas emissions can be found [here](https://arxiv.org/pdf/2204.05149). Since Meta is openly releasing these models, the training energy use and greenhouse gas emissions will not be incurred by others.
413
+
414
+
415
+ ## Training Data
416
+
417
+ **Overview:** Llama 3.1 was pretrained on ~15 trillion tokens of data from publicly available sources. The fine-tuning data includes publicly available instruction datasets, as well as over 25M synthetically generated examples.
418
+
419
+ **Data Freshness:** The pretraining data has a cutoff of December 2023.
420
+
421
+
422
+ ## Benchmark scores
423
+
424
+ In this section, we report the results for Llama 3.1 models on standard automatic benchmarks. For all the evaluations, we use our internal evaluations library.
425
+
426
+ ### Base pretrained models
427
+
428
+
429
+ <table>
430
+ <tr>
431
+ <td><strong>Category</strong>
432
+ </td>
433
+ <td><strong>Benchmark</strong>
434
+ </td>
435
+ <td><strong># Shots</strong>
436
+ </td>
437
+ <td><strong>Metric</strong>
438
+ </td>
439
+ <td><strong>Llama 3 8B</strong>
440
+ </td>
441
+ <td><strong>Llama 3.1 8B</strong>
442
+ </td>
443
+ <td><strong>Llama 3 70B</strong>
444
+ </td>
445
+ <td><strong>Llama 3.1 70B</strong>
446
+ </td>
447
+ <td><strong>Llama 3.1 405B</strong>
448
+ </td>
449
+ </tr>
450
+ <tr>
451
+ <td rowspan="7" >General
452
+ </td>
453
+ <td>MMLU
454
+ </td>
455
+ <td>5
456
+ </td>
457
+ <td>macro_avg/acc_char
458
+ </td>
459
+ <td>66.7
460
+ </td>
461
+ <td>66.7
462
+ </td>
463
+ <td>79.5
464
+ </td>
465
+ <td>79.3
466
+ </td>
467
+ <td>85.2
468
+ </td>
469
+ </tr>
470
+ <tr>
471
+ <td>MMLU-Pro (CoT)
472
+ </td>
473
+ <td>5
474
+ </td>
475
+ <td>macro_avg/acc_char
476
+ </td>
477
+ <td>36.2
478
+ </td>
479
+ <td>37.1
480
+ </td>
481
+ <td>55.0
482
+ </td>
483
+ <td>53.8
484
+ </td>
485
+ <td>61.6
486
+ </td>
487
+ </tr>
488
+ <tr>
489
+ <td>AGIEval English
490
+ </td>
491
+ <td>3-5
492
+ </td>
493
+ <td>average/acc_char
494
+ </td>
495
+ <td>47.1
496
+ </td>
497
+ <td>47.8
498
+ </td>
499
+ <td>63.0
500
+ </td>
501
+ <td>64.6
502
+ </td>
503
+ <td>71.6
504
+ </td>
505
+ </tr>
506
+ <tr>
507
+ <td>CommonSenseQA
508
+ </td>
509
+ <td>7
510
+ </td>
511
+ <td>acc_char
512
+ </td>
513
+ <td>72.6
514
+ </td>
515
+ <td>75.0
516
+ </td>
517
+ <td>83.8
518
+ </td>
519
+ <td>84.1
520
+ </td>
521
+ <td>85.8
522
+ </td>
523
+ </tr>
524
+ <tr>
525
+ <td>Winogrande
526
+ </td>
527
+ <td>5
528
+ </td>
529
+ <td>acc_char
530
+ </td>
531
+ <td>-
532
+ </td>
533
+ <td>60.5
534
+ </td>
535
+ <td>-
536
+ </td>
537
+ <td>83.3
538
+ </td>
539
+ <td>86.7
540
+ </td>
541
+ </tr>
542
+ <tr>
543
+ <td>BIG-Bench Hard (CoT)
544
+ </td>
545
+ <td>3
546
+ </td>
547
+ <td>average/em
548
+ </td>
549
+ <td>61.1
550
+ </td>
551
+ <td>64.2
552
+ </td>
553
+ <td>81.3
554
+ </td>
555
+ <td>81.6
556
+ </td>
557
+ <td>85.9
558
+ </td>
559
+ </tr>
560
+ <tr>
561
+ <td>ARC-Challenge
562
+ </td>
563
+ <td>25
564
+ </td>
565
+ <td>acc_char
566
+ </td>
567
+ <td>79.4
568
+ </td>
569
+ <td>79.7
570
+ </td>
571
+ <td>93.1
572
+ </td>
573
+ <td>92.9
574
+ </td>
575
+ <td>96.1
576
+ </td>
577
+ </tr>
578
+ <tr>
579
+ <td>Knowledge reasoning
580
+ </td>
581
+ <td>TriviaQA-Wiki
582
+ </td>
583
+ <td>5
584
+ </td>
585
+ <td>em
586
+ </td>
587
+ <td>78.5
588
+ </td>
589
+ <td>77.6
590
+ </td>
591
+ <td>89.7
592
+ </td>
593
+ <td>89.8
594
+ </td>
595
+ <td>91.8
596
+ </td>
597
+ </tr>
598
+ <tr>
599
+ <td rowspan="4" >Reading comprehension
600
+ </td>
601
+ <td>SQuAD
602
+ </td>
603
+ <td>1
604
+ </td>
605
+ <td>em
606
+ </td>
607
+ <td>76.4
608
+ </td>
609
+ <td>77.0
610
+ </td>
611
+ <td>85.6
612
+ </td>
613
+ <td>81.8
614
+ </td>
615
+ <td>89.3
616
+ </td>
617
+ </tr>
618
+ <tr>
619
+ <td>QuAC (F1)
620
+ </td>
621
+ <td>1
622
+ </td>
623
+ <td>f1
624
+ </td>
625
+ <td>44.4
626
+ </td>
627
+ <td>44.9
628
+ </td>
629
+ <td>51.1
630
+ </td>
631
+ <td>51.1
632
+ </td>
633
+ <td>53.6
634
+ </td>
635
+ </tr>
636
+ <tr>
637
+ <td>BoolQ
638
+ </td>
639
+ <td>0
640
+ </td>
641
+ <td>acc_char
642
+ </td>
643
+ <td>75.7
644
+ </td>
645
+ <td>75.0
646
+ </td>
647
+ <td>79.0
648
+ </td>
649
+ <td>79.4
650
+ </td>
651
+ <td>80.0
652
+ </td>
653
+ </tr>
654
+ <tr>
655
+ <td>DROP (F1)
656
+ </td>
657
+ <td>3
658
+ </td>
659
+ <td>f1
660
+ </td>
661
+ <td>58.4
662
+ </td>
663
+ <td>59.5
664
+ </td>
665
+ <td>79.7
666
+ </td>
667
+ <td>79.6
668
+ </td>
669
+ <td>84.8
670
+ </td>
671
+ </tr>
672
+ </table>
673
+
674
+
675
+
676
+ ### Instruction tuned models
677
+
678
+
679
+ <table>
680
+ <tr>
681
+ <td><strong>Category</strong>
682
+ </td>
683
+ <td><strong>Benchmark</strong>
684
+ </td>
685
+ <td><strong># Shots</strong>
686
+ </td>
687
+ <td><strong>Metric</strong>
688
+ </td>
689
+ <td><strong>Llama 3 8B Instruct</strong>
690
+ </td>
691
+ <td><strong>Llama 3.1 8B Instruct</strong>
692
+ </td>
693
+ <td><strong>Llama 3 70B Instruct</strong>
694
+ </td>
695
+ <td><strong>Llama 3.1 70B Instruct</strong>
696
+ </td>
697
+ <td><strong>Llama 3.1 405B Instruct</strong>
698
+ </td>
699
+ </tr>
700
+ <tr>
701
+ <td rowspan="4" >General
702
+ </td>
703
+ <td>MMLU
704
+ </td>
705
+ <td>5
706
+ </td>
707
+ <td>macro_avg/acc
708
+ </td>
709
+ <td>68.5
710
+ </td>
711
+ <td>69.4
712
+ </td>
713
+ <td>82.0
714
+ </td>
715
+ <td>83.6
716
+ </td>
717
+ <td>87.3
718
+ </td>
719
+ </tr>
720
+ <tr>
721
+ <td>MMLU (CoT)
722
+ </td>
723
+ <td>0
724
+ </td>
725
+ <td>macro_avg/acc
726
+ </td>
727
+ <td>65.3
728
+ </td>
729
+ <td>73.0
730
+ </td>
731
+ <td>80.9
732
+ </td>
733
+ <td>86.0
734
+ </td>
735
+ <td>88.6
736
+ </td>
737
+ </tr>
738
+ <tr>
739
+ <td>MMLU-Pro (CoT)
740
+ </td>
741
+ <td>5
742
+ </td>
743
+ <td>micro_avg/acc_char
744
+ </td>
745
+ <td>45.5
746
+ </td>
747
+ <td>48.3
748
+ </td>
749
+ <td>63.4
750
+ </td>
751
+ <td>66.4
752
+ </td>
753
+ <td>73.3
754
+ </td>
755
+ </tr>
756
+ <tr>
757
+ <td>IFEval
758
+ </td>
759
+ <td>
760
+ </td>
761
+ <td>
762
+ </td>
763
+ <td>76.8
764
+ </td>
765
+ <td>80.4
766
+ </td>
767
+ <td>82.9
768
+ </td>
769
+ <td>87.5
770
+ </td>
771
+ <td>88.6
772
+ </td>
773
+ </tr>
774
+ <tr>
775
+ <td rowspan="2" >Reasoning
776
+ </td>
777
+ <td>ARC-C
778
+ </td>
779
+ <td>0
780
+ </td>
781
+ <td>acc
782
+ </td>
783
+ <td>82.4
784
+ </td>
785
+ <td>83.4
786
+ </td>
787
+ <td>94.4
788
+ </td>
789
+ <td>94.8
790
+ </td>
791
+ <td>96.9
792
+ </td>
793
+ </tr>
794
+ <tr>
795
+ <td>GPQA
796
+ </td>
797
+ <td>0
798
+ </td>
799
+ <td>em
800
+ </td>
801
+ <td>34.6
802
+ </td>
803
+ <td>30.4
804
+ </td>
805
+ <td>39.5
806
+ </td>
807
+ <td>41.7
808
+ </td>
809
+ <td>50.7
810
+ </td>
811
+ </tr>
812
+ <tr>
813
+ <td rowspan="4" >Code
814
+ </td>
815
+ <td>HumanEval
816
+ </td>
817
+ <td>0
818
+ </td>
819
+ <td>pass@1
820
+ </td>
821
+ <td>60.4
822
+ </td>
823
+ <td>72.6
824
+ </td>
825
+ <td>81.7
826
+ </td>
827
+ <td>80.5
828
+ </td>
829
+ <td>89.0
830
+ </td>
831
+ </tr>
832
+ <tr>
833
+ <td>MBPP ++ base version
834
+ </td>
835
+ <td>0
836
+ </td>
837
+ <td>pass@1
838
+ </td>
839
+ <td>70.6
840
+ </td>
841
+ <td>72.8
842
+ </td>
843
+ <td>82.5
844
+ </td>
845
+ <td>86.0
846
+ </td>
847
+ <td>88.6
848
+ </td>
849
+ </tr>
850
+ <tr>
851
+ <td>Multipl-E HumanEval
852
+ </td>
853
+ <td>0
854
+ </td>
855
+ <td>pass@1
856
+ </td>
857
+ <td>-
858
+ </td>
859
+ <td>50.8
860
+ </td>
861
+ <td>-
862
+ </td>
863
+ <td>65.5
864
+ </td>
865
+ <td>75.2
866
+ </td>
867
+ </tr>
868
+ <tr>
869
+ <td>Multipl-E MBPP
870
+ </td>
871
+ <td>0
872
+ </td>
873
+ <td>pass@1
874
+ </td>
875
+ <td>-
876
+ </td>
877
+ <td>52.4
878
+ </td>
879
+ <td>-
880
+ </td>
881
+ <td>62.0
882
+ </td>
883
+ <td>65.7
884
+ </td>
885
+ </tr>
886
+ <tr>
887
+ <td rowspan="2" >Math
888
+ </td>
889
+ <td>GSM-8K (CoT)
890
+ </td>
891
+ <td>8
892
+ </td>
893
+ <td>em_maj1@1
894
+ </td>
895
+ <td>80.6
896
+ </td>
897
+ <td>84.5
898
+ </td>
899
+ <td>93.0
900
+ </td>
901
+ <td>95.1
902
+ </td>
903
+ <td>96.8
904
+ </td>
905
+ </tr>
906
+ <tr>
907
+ <td>MATH (CoT)
908
+ </td>
909
+ <td>0
910
+ </td>
911
+ <td>final_em
912
+ </td>
913
+ <td>29.1
914
+ </td>
915
+ <td>51.9
916
+ </td>
917
+ <td>51.0
918
+ </td>
919
+ <td>68.0
920
+ </td>
921
+ <td>73.8
922
+ </td>
923
+ </tr>
924
+ <tr>
925
+ <td rowspan="4" >Tool Use
926
+ </td>
927
+ <td>API-Bank
928
+ </td>
929
+ <td>0
930
+ </td>
931
+ <td>acc
932
+ </td>
933
+ <td>48.3
934
+ </td>
935
+ <td>82.6
936
+ </td>
937
+ <td>85.1
938
+ </td>
939
+ <td>90.0
940
+ </td>
941
+ <td>92.0
942
+ </td>
943
+ </tr>
944
+ <tr>
945
+ <td>BFCL
946
+ </td>
947
+ <td>0
948
+ </td>
949
+ <td>acc
950
+ </td>
951
+ <td>60.3
952
+ </td>
953
+ <td>76.1
954
+ </td>
955
+ <td>83.0
956
+ </td>
957
+ <td>84.8
958
+ </td>
959
+ <td>88.5
960
+ </td>
961
+ </tr>
962
+ <tr>
963
+ <td>Gorilla Benchmark API Bench
964
+ </td>
965
+ <td>0
966
+ </td>
967
+ <td>acc
968
+ </td>
969
+ <td>1.7
970
+ </td>
971
+ <td>8.2
972
+ </td>
973
+ <td>14.7
974
+ </td>
975
+ <td>29.7
976
+ </td>
977
+ <td>35.3
978
+ </td>
979
+ </tr>
980
+ <tr>
981
+ <td>Nexus (0-shot)
982
+ </td>
983
+ <td>0
984
+ </td>
985
+ <td>macro_avg/acc
986
+ </td>
987
+ <td>18.1
988
+ </td>
989
+ <td>38.5
990
+ </td>
991
+ <td>47.8
992
+ </td>
993
+ <td>56.7
994
+ </td>
995
+ <td>58.7
996
+ </td>
997
+ </tr>
998
+ <tr>
999
+ <td>Multilingual
1000
+ </td>
1001
+ <td>Multilingual MGSM (CoT)
1002
+ </td>
1003
+ <td>0
1004
+ </td>
1005
+ <td>em
1006
+ </td>
1007
+ <td>-
1008
+ </td>
1009
+ <td>68.9
1010
+ </td>
1011
+ <td>-
1012
+ </td>
1013
+ <td>86.9
1014
+ </td>
1015
+ <td>91.6
1016
+ </td>
1017
+ </tr>
1018
+ </table>
1019
+
1020
+ #### Multilingual benchmarks
1021
+
1022
+ <table>
1023
+ <tr>
1024
+ <td><strong>Category</strong>
1025
+ </td>
1026
+ <td><strong>Benchmark</strong>
1027
+ </td>
1028
+ <td><strong>Language</strong>
1029
+ </td>
1030
+ <td><strong>Llama 3.1 8B</strong>
1031
+ </td>
1032
+ <td><strong>Llama 3.1 70B</strong>
1033
+ </td>
1034
+ <td><strong>Llama 3.1 405B</strong>
1035
+ </td>
1036
+ </tr>
1037
+ <tr>
1038
+ <td rowspan="9" ><strong>General</strong>
1039
+ </td>
1040
+ <td rowspan="9" ><strong>MMLU (5-shot, macro_avg/acc)</strong>
1041
+ </td>
1042
+ <td>Portuguese
1043
+ </td>
1044
+ <td>62.12
1045
+ </td>
1046
+ <td>80.13
1047
+ </td>
1048
+ <td>84.95
1049
+ </td>
1050
+ </tr>
1051
+ <tr>
1052
+ <td>Spanish
1053
+ </td>
1054
+ <td>62.45
1055
+ </td>
1056
+ <td>80.05
1057
+ </td>
1058
+ <td>85.08
1059
+ </td>
1060
+ </tr>
1061
+ <tr>
1062
+ <td>Italian
1063
+ </td>
1064
+ <td>61.63
1065
+ </td>
1066
+ <td>80.4
1067
+ </td>
1068
+ <td>85.04
1069
+ </td>
1070
+ </tr>
1071
+ <tr>
1072
+ <td>German
1073
+ </td>
1074
+ <td>60.59
1075
+ </td>
1076
+ <td>79.27
1077
+ </td>
1078
+ <td>84.36
1079
+ </td>
1080
+ </tr>
1081
+ <tr>
1082
+ <td>French
1083
+ </td>
1084
+ <td>62.34
1085
+ </td>
1086
+ <td>79.82
1087
+ </td>
1088
+ <td>84.66
1089
+ </td>
1090
+ </tr>
1091
+ <tr>
1092
+ <td>Hindi
1093
+ </td>
1094
+ <td>50.88
1095
+ </td>
1096
+ <td>74.52
1097
+ </td>
1098
+ <td>80.31
1099
+ </td>
1100
+ </tr>
1101
+ <tr>
1102
+ <td>Thai
1103
+ </td>
1104
+ <td>50.32
1105
+ </td>
1106
+ <td>72.95
1107
+ </td>
1108
+ <td>78.21
1109
+ </td>
1110
+ </tr>
1111
+ </table>
1112
+
1113
+
1114
+
1115
+ ## Responsibility & Safety
1116
+
1117
+ As part of our Responsible release approach, we followed a three-pronged strategy to managing trust & safety risks:
1118
+
1119
+
1120
+
1121
+ * Enable developers to deploy helpful, safe and flexible experiences for their target audience and for the use cases supported by Llama.
1122
+ * Protect developers against adversarial users aiming to exploit Llama capabilities to potentially cause harm.
1123
+ * Provide protections for the community to help prevent the misuse of our models.
1124
+
1125
+
1126
+ ### Responsible deployment
1127
+
1128
+ Llama is a foundational technology designed to be used in a variety of use cases, examples on how Meta’s Llama models have been responsibly deployed can be found in our [Community Stories webpage](https://llama.meta.com/community-stories/). Our approach is to build the most helpful models enabling the world to benefit from the technology power, by aligning our model safety for the generic use cases addressing a standard set of harms. Developers are then in the driver seat to tailor safety for their use case, defining their own policy and deploying the models with the necessary safeguards in their Llama systems. Llama 3.1 was developed following the best practices outlined in our Responsible Use Guide, you can refer to the [Responsible Use Guide](https://llama.meta.com/responsible-use-guide/) to learn more.
1129
+
1130
+
1131
+ #### Llama 3.1 instruct
1132
+
1133
+ Our main objectives for conducting safety fine-tuning are to provide the research community with a valuable resource for studying the robustness of safety fine-tuning, as well as to offer developers a readily available, safe, and powerful model for various applications to reduce the developer workload to deploy safe AI systems. For more details on the safety mitigations implemented please read the Llama 3 paper.
1134
+
1135
+ **Fine-tuning data**
1136
+
1137
+ We employ a multi-faceted approach to data collection, combining human-generated data from our vendors with synthetic data to mitigate potential safety risks. We’ve developed many large language model (LLM)-based classifiers that enable us to thoughtfully select high-quality prompts and responses, enhancing data quality control.
1138
+
1139
+ **Refusals and Tone**
1140
+
1141
+ Building on the work we started with Llama 3, we put a great emphasis on model refusals to benign prompts as well as refusal tone. We included both borderline and adversarial prompts in our safety data strategy, and modified our safety data responses to follow tone guidelines.
1142
+
1143
+
1144
+ #### Llama 3.1 systems
1145
+
1146
+ **Large language models, including Llama 3.1, are not designed to be deployed in isolation but instead should be deployed as part of an overall AI system with additional safety guardrails as required.** Developers are expected to deploy system safeguards when building agentic systems. Safeguards are key to achieve the right helpfulness-safety alignment as well as mitigating safety and security risks inherent to the system and any integration of the model or system with external tools.
1147
+
1148
+ As part of our responsible release approach, we provide the community with [safeguards](https://llama.meta.com/trust-and-safety/) that developers should deploy with Llama models or other LLMs, including Llama Guard 3, Prompt Guard and Code Shield. All our [reference implementations](https://github.com/meta-llama/llama-agentic-system) demos contain these safeguards by default so developers can benefit from system-level safety out-of-the-box.
1149
+
1150
+
1151
+ #### New capabilities
1152
+
1153
+ Note that this release introduces new capabilities, including a longer context window, multilingual inputs and outputs and possible integrations by developers with third party tools. Building with these new capabilities requires specific considerations in addition to the best practices that generally apply across all Generative AI use cases.
1154
+
1155
+ **Tool-use**: Just like in standard software development, developers are responsible for the integration of the LLM with the tools and services of their choice. They should define a clear policy for their use case and assess the integrity of the third party services they use to be aware of the safety and security limitations when using this capability. Refer to the Responsible Use Guide for best practices on the safe deployment of the third party safeguards.
1156
+
1157
+ **Multilinguality**: Llama 3.1 supports 7 languages in addition to English: French, German, Hindi, Italian, Portuguese, Spanish, and Thai. Llama may be able to output text in other languages than those that meet performance thresholds for safety and helpfulness. We strongly discourage developers from using this model to converse in non-supported languages without implementing finetuning and system controls in alignment with their policies and the best practices shared in the Responsible Use Guide.
1158
+
1159
+
1160
+ ### Evaluations
1161
+
1162
+ We evaluated Llama models for common use cases as well as specific capabilities. Common use cases evaluations measure safety risks of systems for most commonly built applications including chat bot, coding assistant, tool calls. We built dedicated, adversarial evaluation datasets and evaluated systems composed of Llama models and Llama Guard 3 to filter input prompt and output response. It is important to evaluate applications in context, and we recommend building dedicated evaluation dataset for your use case. Prompt Guard and Code Shield are also available if relevant to the application.
1163
+
1164
+ Capability evaluations measure vulnerabilities of Llama models inherent to specific capabilities, for which were crafted dedicated benchmarks including long context, multilingual, tools calls, coding or memorization.
1165
+
1166
+ **Red teaming**
1167
+
1168
+ For both scenarios, we conducted recurring red teaming exercises with the goal of discovering risks via adversarial prompting and we used the learnings to improve our benchmarks and safety tuning datasets.
1169
+
1170
+ We partnered early with subject-matter experts in critical risk areas to understand the nature of these real-world harms and how such models may lead to unintended harm for society. Based on these conversations, we derived a set of adversarial goals for the red team to attempt to achieve, such as extracting harmful information or reprogramming the model to act in a potentially harmful capacity. The red team consisted of experts in cybersecurity, adversarial machine learning, responsible AI, and integrity in addition to multilingual content specialists with background in integrity issues in specific geographic markets.
1171
+
1172
+
1173
+ ### Critical and other risks
1174
+
1175
+ We specifically focused our efforts on mitigating the following critical risk areas:
1176
+
1177
+ **1- CBRNE (Chemical, Biological, Radiological, Nuclear, and Explosive materials) helpfulness**
1178
+
1179
+ To assess risks related to proliferation of chemical and biological weapons, we performed uplift testing designed to assess whether use of Llama 3.1 models could meaningfully increase the capabilities of malicious actors to plan or carry out attacks using these types of weapons.
1180
+
1181
+
1182
+ **2. Child Safety**
1183
+
1184
+ Child Safety risk assessments were conducted using a team of experts, to assess the model’s capability to produce outputs that could result in Child Safety risks and inform on any necessary and appropriate risk mitigations via fine tuning. We leveraged those expert red teaming sessions to expand the coverage of our evaluation benchmarks through Llama 3 model development. For Llama 3, we conducted new in-depth sessions using objective based methodologies to assess the model risks along multiple attack vectors including the additional languages Llama 3 is trained on. We also partnered with content specialists to perform red teaming exercises assessing potentially violating content while taking account of market specific nuances or experiences.
1185
+
1186
+ **3. Cyber attack enablement**
1187
+
1188
+ Our cyber attack uplift study investigated whether LLMs can enhance human capabilities in hacking tasks, both in terms of skill level and speed.
1189
+
1190
+ Our attack automation study focused on evaluating the capabilities of LLMs when used as autonomous agents in cyber offensive operations, specifically in the context of ransomware attacks. This evaluation was distinct from previous studies that considered LLMs as interactive assistants. The primary objective was to assess whether these models could effectively function as independent agents in executing complex cyber-attacks without human intervention.
1191
+
1192
+ Our study of Llama-3.1-405B’s social engineering uplift for cyber attackers was conducted to assess the effectiveness of AI models in aiding cyber threat actors in spear phishing campaigns. Please read our Llama 3.1 Cyber security whitepaper to learn more.
1193
+
1194
+
1195
+ ### Community
1196
+
1197
+ Generative AI safety requires expertise and tooling, and we believe in the strength of the open community to accelerate its progress. We are active members of open consortiums, including the AI Alliance, Partnership on AI and MLCommons, actively contributing to safety standardization and transparency. We encourage the community to adopt taxonomies like the MLCommons Proof of Concept evaluation to facilitate collaboration and transparency on safety and content evaluations. Our Purple Llama tools are open sourced for the community to use and widely distributed across ecosystem partners including cloud service providers. We encourage community contributions to our [Github repository](https://github.com/meta-llama/PurpleLlama).
1198
+
1199
+ We also set up the [Llama Impact Grants](https://llama.meta.com/llama-impact-grants/) program to identify and support the most compelling applications of Meta’s Llama model for societal benefit across three categories: education, climate and open innovation. The 20 finalists from the hundreds of applications can be found [here](https://llama.meta.com/llama-impact-grants/#finalists).
1200
+
1201
+ Finally, we put in place a set of resources including an [output reporting mechanism](https://developers.facebook.com/llama_output_feedback) and [bug bounty program](https://www.facebook.com/whitehat) to continuously improve the Llama technology with the help of the community.
1202
+
1203
+
1204
+ ## Ethical Considerations and Limitations
1205
+
1206
+ The core values of Llama 3.1 are openness, inclusivity and helpfulness. It is meant to serve everyone, and to work for a wide range of use cases. It is thus designed to be accessible to people across many different backgrounds, experiences and perspectives. Llama 3.1 addresses users and their needs as they are, without insertion unnecessary judgment or normativity, while reflecting the understanding that even content that may appear problematic in some cases can serve valuable purposes in others. It respects the dignity and autonomy of all users, especially in terms of the values of free thought and expression that power innovation and progress.
1207
+
1208
+ But Llama 3.1 is a new technology, and like any new technology, there are risks associated with its use. Testing conducted to date has not covered, nor could it cover, all scenarios. For these reasons, as with all LLMs, Llama 3.1’s potential outputs cannot be predicted in advance, and the model may in some instances produce inaccurate, biased or other objectionable responses to user prompts. Therefore, before deploying any applications of Llama 3.1 models, developers should perform safety testing and tuning tailored to their specific applications of the model. Please refer to available resources including our [Responsible Use Guide](https://llama.meta.com/responsible-use-guide), [Trust and Safety](https://llama.meta.com/trust-and-safety/) solutions, and other [resources](https://llama.meta.com/docs/get-started/) to learn more about responsible development.
USE_POLICY.md ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Llama 3.1 Acceptable Use Policy
2
+
3
+ Meta is committed to promoting safe and fair use of its tools and features, including Llama 3.1. If you
4
+ access or use Llama 3.1, you agree to this Acceptable Use Policy (“Policy”). The most recent copy of
5
+ this policy can be found at [https://llama.meta.com/llama3_1/use-policy](https://llama.meta.com/llama3_1/use-policy)
6
+
7
+ ## Prohibited Uses
8
+
9
+ We want everyone to use Llama 3.1 safely and responsibly. You agree you will not use, or allow
10
+ others to use, Llama 3.1 to:
11
+
12
+ 1. Violate the law or others’ rights, including to:
13
+ 1. Engage in, promote, generate, contribute to, encourage, plan, incite, or further illegal or unlawful activity or content, such as:
14
+ 1. Violence or terrorism
15
+ 2. Exploitation or harm to children, including the solicitation, creation, acquisition, or dissemination of child exploitative content or failure to report Child Sexual Abuse Material
16
+ 3. Human trafficking, exploitation, and sexual violence
17
+ 4. The illegal distribution of information or materials to minors, including obscene materials, or failure to employ legally required age-gating in connection with such information or materials.
18
+ 5. Sexual solicitation
19
+ 6. Any other criminal activity
20
+ 3. Engage in, promote, incite, or facilitate the harassment, abuse, threatening, or bullying of individuals or groups of individuals
21
+ 4. Engage in, promote, incite, or facilitate discrimination or other unlawful or harmful conduct in the provision of employment, employment benefits, credit, housing, other economic benefits, or other essential goods and services
22
+ 5. Engage in the unauthorized or unlicensed practice of any profession including, but not limited to, financial, legal, medical/health, or related professional practices
23
+ 6. Collect, process, disclose, generate, or infer health, demographic, or other sensitive personal or private information about individuals without rights and consents required by applicable laws
24
+ 7. Engage in or facilitate any action or generate any content that infringes, misappropriates, or otherwise violates any third-party rights, including the outputs or results of any products or services using the Llama Materials
25
+ 8. Create, generate, or facilitate the creation of malicious code, malware, computer viruses or do anything else that could disable, overburden, interfere with or impair the proper working, integrity, operation or appearance of a website or computer system
26
+
27
+ 2. Engage in, promote, incite, facilitate, or assist in the planning or development of activities that present a risk of death or bodily harm to individuals, including use of Llama 3.1 related to the following:
28
+ 1. Military, warfare, nuclear industries or applications, espionage, use for materials or activities that are subject to the International Traffic Arms Regulations (ITAR) maintained by the United States Department of State
29
+ 2. Guns and illegal weapons (including weapon development)
30
+ 3. Illegal drugs and regulated/controlled substances
31
+ 4. Operation of critical infrastructure, transportation technologies, or heavy machinery
32
+ 5. Self-harm or harm to others, including suicide, cutting, and eating disorders
33
+ 6. Any content intended to incite or promote violence, abuse, or any infliction of bodily harm to an individual
34
+
35
+ 3. Intentionally deceive or mislead others, including use of Llama 3.1 related to the following:
36
+ 1. Generating, promoting, or furthering fraud or the creation or promotion of disinformation
37
+ 2. Generating, promoting, or furthering defamatory content, including the creation of defamatory statements, images, or other content
38
+ 3. Generating, promoting, or further distributing spam
39
+ 4. Impersonating another individual without consent, authorization, or legal right
40
+ 5. Representing that the use of Llama 3.1 or outputs are human-generated
41
+ 6. Generating or facilitating false online engagement, including fake reviews and other means of fake online engagement
42
+
43
+ 4. Fail to appropriately disclose to end users any known dangers of your AI system
44
+
45
+ Please report any violation of this Policy, software “bug,” or other problems that could lead to a violation
46
+ of this Policy through one of the following means:
47
+
48
+ * Reporting issues with the model: [https://github.com/meta-llama/llama-models/issues](https://github.com/meta-llama/llama-models/issues)
49
+ * Reporting risky content generated by the model: developers.facebook.com/llama_output_feedback
50
+ * Reporting bugs and security concerns: facebook.com/whitehat/info
51
+ * Reporting violations of the Acceptable Use Policy or unlicensed uses of Llama 3.1: [email protected]