leafspark commited on
Commit
ed923a9
β€’
1 Parent(s): e0ae354

Add benchmarks

Browse files
Files changed (1) hide show
  1. README.md +29 -22
README.md CHANGED
@@ -5,12 +5,14 @@ language:
5
  - en
6
  - fr
7
  - es
 
 
8
  library_name: transformers
9
  tags:
10
  - moe
11
  - text-generation-inference
12
  ---
13
- # Mixtral-8x22b
14
  New MoE model by MistralAI
15
 
16
  ## Model Details:
@@ -19,6 +21,17 @@ New MoE model by MistralAI
19
  - 56 layers
20
  - 8 experts
21
 
 
 
 
 
 
 
 
 
 
 
 
22
  The model is split into 27 parts, from the original torrent.
23
 
24
  Magnet link and checksum: [https://twitter.com/mistralai/status/1777869263778291896](https://twitter.com/mistralai/status/1777869263778291896)
@@ -34,7 +47,6 @@ Make sure you have Python 2 or 3 installed (HuggingFace libraries not required):
34
  python merge.py
35
  ```
36
  This should take approximately 2 hours, you will be left with a 274GB file.
37
-
38
  Check the MD5 hash of consolidated.safetensors:
39
  ```
40
  3816cd2c4f827b4b868bc6481d5d3ba2
@@ -43,6 +55,7 @@ Check the MD5 hash of consolidated.safetensors:
43
  That's it! Now you have the complete torrent download on your computer.
44
 
45
  ## Credit to:
 
46
 
47
  β–„β–„β–„β–‘β–‘
48
  β–„β–„β–„β–„β–„β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‘β–‘β–‘β–‘
@@ -84,23 +97,17 @@ That's it! Now you have the complete torrent download on your computer.
84
  β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘
85
  β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘
86
  β–‘β–‘β–‘β–‘β–‘
87
-
88
- ╓────────────────────────────────────────────────────────────────────────────╖
89
-
90
- β•‘ β•‘
91
- β•‘ Β·Β· Released by the Mistral AI team Β·Β· β•‘
92
-
93
- β•‘ Albert Jiang, Alexandre Sablayrolles, Alexis Tacnet, Antoine Roux, β•‘
94
- β•‘ Arthur Mensch, Audrey Herblin-Stoop, Baptiste Bout, Baudouin de Monicault, β•‘
95
- β•‘ Blanche Savary, Bam4d, Caroline Feldman, Devendra Singh Chaplot, β•‘
96
- β•‘ Diego de las Casas, Eleonore Arcelin, Emma Bou Hanna, Etienne Metzger, β•‘
97
- β•‘ Gianna Lengyel, Guillaume Bour, Guillaume Lample, Harizo Rajaona, β•‘
98
- β•‘ Jean-Malo Delignon, Jia Li, Justus Murke, Louis Martin, Louis Ternon β•‘
99
- β•‘ Lucile Saulnier, LΓ©lio Renard Lavaud, Margaret Jennings, Marie Pellat, β•‘
100
- β•‘ Marie Torelli, Marie-Anne Lachaux, Nicolas Schuhl, Patrick von Platen, β•‘
101
- β•‘ Pierre Stock, Sandeep Subramanian, Sophia Yang, Szymon Antoniak, β•‘
102
- β•‘ Teven Le Scao, Thibaut Lavril, TimothΓ©e Lacroix, ThΓ©ophile Gervet, β•‘
103
- β•‘ Thomas Wang, Valera Nemychnikova, William El Sayed, William Marshall β•‘
104
- β•‘ β•‘
105
-
106
- β•™β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β•œ
 
5
  - en
6
  - fr
7
  - es
8
+ - it
9
+ - de
10
  library_name: transformers
11
  tags:
12
  - moe
13
  - text-generation-inference
14
  ---
15
+ # Mixtral-8x22B-v0.1
16
  New MoE model by MistralAI
17
 
18
  ## Model Details:
 
21
  - 56 layers
22
  - 8 experts
23
 
24
+ ## Benchmarks
25
+ ```
26
+ ARC C (25-shot): 70.5
27
+ Hellaswag (10-shot): 88.9
28
+ MMLU (5-shot): 77.3
29
+ TruthfulQA: 52.3
30
+ Winogrande (5-shot): 85.2
31
+ GSM8K (5-shot): 76.5
32
+
33
+ Source: https://huggingface.co/mistral-community/Mixtral-8x22B-v0.1/discussions/4#6616c393b8d25135997cdd45
34
+ ```
35
  The model is split into 27 parts, from the original torrent.
36
 
37
  Magnet link and checksum: [https://twitter.com/mistralai/status/1777869263778291896](https://twitter.com/mistralai/status/1777869263778291896)
 
47
  python merge.py
48
  ```
49
  This should take approximately 2 hours, you will be left with a 274GB file.
 
50
  Check the MD5 hash of consolidated.safetensors:
51
  ```
52
  3816cd2c4f827b4b868bc6481d5d3ba2
 
55
  That's it! Now you have the complete torrent download on your computer.
56
 
57
  ## Credit to:
58
+ ```
59
 
60
  β–„β–„β–„β–‘β–‘
61
  β–„β–„β–„β–„β–„β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‘β–‘β–‘β–‘
 
97
  β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘
98
  β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘β–‘
99
  β–‘β–‘β–‘β–‘β–‘
100
+ ```
101
+ ## Released by the Mistral AI team:
102
+ Albert Jiang, Alexandre Sablayrolles, Alexis Tacnet, Antoine Roux,
103
+ Arthur Mensch, Audrey Herblin-Stoop, Baptiste Bout, Baudouin de Monicault,
104
+ Blanche Savary, Bam4d, Caroline Feldman, Devendra Singh Chaplot,
105
+ Diego de las Casas, Eleonore Arcelin, Emma Bou Hanna, Etienne Metzger,
106
+ Gianna Lengyel, Guillaume Bour, Guillaume Lample, Harizo Rajaona,
107
+ Jean-Malo Delignon, Jia Li, Justus Murke, Louis Martin, Louis Ternon
108
+ Lucile Saulnier, LΓ©lio Renard Lavaud, Margaret Jennings, Marie Pellat,
109
+ Marie Torelli, Marie-Anne Lachaux, Nicolas Schuhl, Patrick von Platen,
110
+ Pierre Stock, Sandeep Subramanian, Sophia Yang, Szymon Antoniak,
111
+ Teven Le Scao, Thibaut Lavril, TimothΓ©e Lacroix, ThΓ©ophile Gervet,
112
+ Thomas Wang, Valera Nemychnikova, William El Sayed, William Marshall
113
+