Hugging Face's logo

Mathoctopus
/
Parallel_7B

Text Generation
Transformers
PyTorch
llama
text-generation-inference
Inference Endpoints
Model card Files Community
Nuo97 commited on
Commit
9e07b26
•
1 Parent(s): a0a1d18

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +118 -1
README.md CHANGED
@@ -1,3 +1,120 @@
1
  ---
2
  license: apache-2.0
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: apache-2.0
3
+ ---
4
+
5
+ ### Introduction
6
+
7
+ We introduce 🐙 MathOctopus, a series of open-source large language models (LLMs) specifically tailored for multilingual math problem-solving. The MathOctopus models are trained on 🤗 MGSM8KInstruct Dataset, encompassing ten distinct languages.
8
+ MathOctopus notably outperforms conventional open-source LLMs and exhibits superiority over ChatGPT in few-shot scenarios.
9
+
10
+ ### Datasets
11
+
12
+ #### **MGSM8KInstruct**
13
+
14
+ | Training Dataset | En | Sw | Zh | Bn | De | Es | Fr | Ja | Ru | Th | Overall |
15
+ |:----------------------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|
16
+ | MGSM8KInstruct | 7473 | 7472 | 7466 | 6539 | 7466 | 7470 | 7469 | 7471 | 7361 | 7473 | **73.6K** |
17
+
18
+
19
+ #### **MSVAMP**
20
+
21
+ | Test Dataset | En | Sw | Zh | Bn | De | Es | Fr | Ja | Ru | Th | Overall |
22
+ |:----------------------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|
23
+ | MSVAMP | 1000 | 1000 | 1000 | 1000 | 1000 | 1000 | 1000 | 1000 | 1000 | 1000 | **10K** |
24
+
25
+ #### Usage
26
+
27
+ Our dataset and models are all available at Huggingface.
28
+
29
+ 🤗 [MathInstruct Dataset](https://huggingface.co/datasets/Mathoctopus/GSM8KInstruct_Parallel)
30
+
31
+
32
+ Or you can directly download them from
33
+
34
+ ## Models
35
+
36
+ | Base Model: LLama | Parallel-Training | Cross-Training |
37
+ |----|---------------------------------------------------------------|---------------------------------------------------------------------------|
38
+ | 7B-LLaMA 2 | 🐙 [MathOctopus-Parallel-7B](https://huggingface.co/Mathoctopus/Parallel_7B) | 🐙 [MathOctopus-Cross-7B](https://huggingface.co/Mathoctopus/Cross_7B) |
39
+ || 🐙[MathOctopus-Parallel-xRFT-7B](https://huggingface.co/Mathoctopus/Parallel_xRFT_7B)|🐙[MathOctopus-Cross-xRFT-7B](https://huggingface.co/Mathoctopus/Cross_xRFT_7B)|
40
+ | 13B-LLaMA 2 | 🐙 [MathOctopus-Parallel-13B] | 🐙 [MathOctopus-Cross-13B] |
41
+ || 🐙[MathOctopus-Parallel-xRFT-13B](https://huggingface.co/Mathoctopus/Parallel_xRFT_13B/tree/main)|🐙[MathOctopus-Cross-xRFT-13B]()|
42
+ | 33B-LLaMA 1 | 🐙 [MathOctopus-Parallel-33B] | 🐙 [MathOctopus-Cross-33B] |
43
+ | 70B-LLaMA 2 | Coming soon! | Coming Soon! |
44
+
45
+ *-Parallel refers to our model trained with the parallel-training strategy.
46
+
47
+ *-Cross refers to our model trained with cross-training strategy.
48
+
49
+ *-xRFT means we train the model with multilingual rejection sampling.
50
+
51
+ ### **Overall Results on MGSM**
52
+
53
+ | 7B Model | En | Sw | Zh | Bn | De | Es | Fr | Ja | Ru | Th | Overall |
54
+ |:--------------------------------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|
55
+ | MathOctupos<sup>C</sup> | 52.0 | 23.6 | 31.6 | 18.8 | 38.0 | 39.2 | 36.4 | 27.2 | 33.6 | 21.6 | 32.2 |
56
+ | **xRFT**-MathOctupos<sup>C</sup>| 51.2 | 24.0 | 33.2 | 18.8 | 36.0 | 41.2 | 37.6 | 29.6 | 36.4 | 25.2 | 33.3 |
57
+ | MathOctupos<sup>P</sup>-LoRA | 30.4 | 15.2 | 23.6 | 10.4 | 22.8 | 24.8 | 26.4 | 18.0 | 22.0 | 14.8 | 20.8 |
58
+ | MathOctupos<sup>P</sup> | 52.4 | 39.2 | 38.4 | 28.8 | 44.8 | 42.4 | 43.6 | 36.0 | 39.6 | 34.4 | 40.0 |
59
+ | **xRFT**-MathOctupos<sup>P</sup>| 54.8 | 38.4 | 45.2 | 33.2 | 43.6 | 45.2 | 38.0 | 35.6 | 48.4 | 36.4 | 41.9 |
60
+ <p></p >
61
+
62
+ | 13B Model | En | Sw | Zh | Bn | De | Es | Fr | Ja | Ru | Th | Overall |
63
+ |:--------------------------------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|
64
+ | MathOctupos<sup>C</sup> | 56.4 | 27.2 | 39.2 | 24.0 | 47.6 | 49.6 | 47.6 | 40.4 | 42.0 | 24.8 | 39.9 |
65
+ | **xRFT**-MathOctupos<sup>C</sup>| 53.6 | 28.0 | 45.2 | 21.2 | 48.0 | 46.4 | 46.0 | 35.2 | 45.6 | 28.8 | 39.8 |
66
+ | MathOctupos<sup>P</sup> | 53.2 | 42.8 | 48.8 | 35.2 | 44.4 | 48.0 | 48.4 | 43.2 | 47.6 | 46.8 | 45.8 |
67
+ | **xRFT**-MathOctupos<sup>P</sup>| 51.6 | 46.0 | 51.2 | 42.0 | 49.2 | 53.2 | 49.6 | 39.6 | 47.6 | 46.0 | 47.6 |
68
+ <p></p >
69
+
70
+ | 30-34B Model | En | Sw | Zh | Bn | De | Es | Fr | Ja | Ru | Th | Overall |
71
+ |:--------------------------------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|
72
+ | MathOctupos<sup>C</sup> | 55.6 | 24.4 | 36.0 | 19.2 | 40.4 | 51.2 | 44.4 | 27.2 | 37.2 | 21.6 | 35.7 |
73
+ | **xRFT**-MathOctupos<sup>C</sup>| 53.6 | 27.6 | 34.4 | 19.2 | 47.2 | 47.6 | 44.8 | 30.8 | 38.8 | 22.8 | 36.7 |
74
+ | MathOctupos<sup>P</sup> | 56.4 | 46.8 | 52.0 | 35.2 | 47.2 | 53.2 | 48.0 | 39.2 | 45.6 | 41.2 | 46.5 |
75
+ | **xRFT**-MathOctupos<sup>P</sup>| 51.6 | 47.2 | 52.4 | 37.6 | 51.2 | 52.8 | 44.4 | 41.6 | 50.0 | 47.6 | 47.6 |
76
+
77
+
78
+ ### **Overall Results on MSVAMP**
79
+
80
+ | 7B Model | En | Sw | Zh | Bn | De | Es | Fr | Ja | Ru | Th | Overall |
81
+ |:--------------------------------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|
82
+ | MathOctupos<sup>C</sup> | 49.2 | 36.6 | 43.6 | 30.2 | 48.6 | 46.8 | 46.4 | 42.5 | 46.7 | 34.0 | 42.5 |
83
+ | **xRFT**-MathOctupos<sup>C</sup>| 49.9 | 37.7 | 43.3 | 32.9 | 46.5 | 47.6 | 47.3 | 42.7 | 46.6 | 36.2 | 43.1 |
84
+ | MathOctupos<sup>P</sup>-LoRA | 30.4 | 15.2 | 23.6 | 10.4 | 22.8 | 24.8 | 26.4 | 18.0 | 22.0 | 14.8 | 20.8 |
85
+ | MathOctupos<sup>P</sup> | 46.5 | 40.1 | 42.5 | 29.1 | 43.5 | 45.4 | 46.0 | 42.5 | 45.4 | 35.7 | 41.7 |
86
+ | **xRFT**-MathOctupos<sup>P</sup>| 46.8 | 42.3 | 43.2 | 32.8 | 43.1 | 44.5 | 45.3 | 43.2 | 42.1 | 40.5 | 42.4 |
87
+ <p></p >
88
+
89
+ | 13B Model | En | Sw | Zh | Bn | De | Es | Fr | Ja | Ru | Th | Overall |
90
+ |:--------------------------------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|
91
+ | MathOctupos<sup>C</sup> | 56.6 | 40.4 | 49.0 | 30.3 | 50.9 | 54.2 | 54.7 | 46.3 | 52.4 | 35.7 | 47.1 |
92
+ | **xRFT**-MathOctupos<sup>C</sup>| 52.9 | 41.9 | 49.2 | 34.1 | 50.5 | 52.8 | 51.5 | 45.8 | 50.2 | 35.7 | 46.5 |
93
+ | MathOctupos<sup>P</sup> | 50.7 | 43.4 | 42.6 | 31.8 | 48.4 | 49.4 | 50.6 | 41.1 | 46.9 | 39.3 | 44.4 |
94
+ | **xRFT**-MathOctupos<sup>P</sup>| 44.6 | 43.4 | 46.4 | 34.2 | 47.7 | 48.2 | 49.9 | 43.1 | 48.2 | 39.5 | 44.5 |
95
+ <p></p >
96
+
97
+ | 30-34B Model | En | Sw | Zh | Bn | De | Es | Fr | Ja | Ru | Th | Overall |
98
+ |:--------------------------------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|
99
+ | MathOctupos<sup>C</sup> | 51.5 | 42.1 | 46.2 | 23.2 | 50.5 | 52.1 | 52.9 | 42.2 | 50.5 | 33.4 | 44.5 |
100
+ | **xRFT**-MathOctupos<sup>C</sup>| 48.1 | 42.8 | 43.6 | 23.3 | 48.7 | 50.0 | 48.9 | 43.4 | 44.6 | 35.5 | 42.9 |
101
+ | MathOctupos<sup>P</sup> | 56.4 | 46.8 | 52.0 | 35.2 | 47.2 | 53.2 | 48.0 | 39.2 | 45.6 | 41.2 | 46.5 |
102
+ | **xRFT**-MathOctupos<sup>P</sup>| 48.0 | 42.3 | 46.1 | 36.2 | 47.5 | 48.5 | 48.3 | 45.8 | 47.2 | 41.2 | 45.1 |
103
+
104
+
105
+ ### **MathOctupos in English**
106
+
107
+ | Models | GSM8K | SVAMP |
108
+ |:--------------------------------|:--------|:--------|
109
+ | LLaMA 2-7B | 42.4 | 38.3 |
110
+ | MathOctupos<sup>P</sup>-7B | 49.3 | 46.8 |
111
+ | MathOctupos<sup>C</sup>-7B | 50.8 | 49.3 |
112
+ | LLaMA 2-13B | 51.0 | 50.9 |
113
+ | MathOctupos<sup>P</sup>-13B | 55.5 | 52.1 |
114
+ | MathOctupos<sup>C</sup>-13B | 56.6 | 56.6 |
115
+ | LLaMA 1-33B | 50.0 | 49.0 |
116
+ | MathOctupos<sup>P</sup>-33B | 56.0 | 52.5 |
117
+ | MathOctupos<sup>C</sup>-33B | 53.7 | 51.5 |
118
+
119
+ ## Intended Uses
120
+ These models are trained for research purposes. They are designed to solve multilingual math problems. They can be used in educational software, tutoring systems, or any application where a solution to a math problem is needed.