Commit
β’
b6e8c8b
1
Parent(s):
0309b18
Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,123 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
datasets:
|
4 |
+
- Mathoctopus/GSM8KInstruct_Parallel
|
5 |
+
language:
|
6 |
+
- en
|
7 |
+
- es
|
8 |
+
- zh
|
9 |
+
- de
|
10 |
+
- ru
|
11 |
+
- th
|
12 |
+
- sw
|
13 |
+
- ja
|
14 |
+
- fr
|
15 |
+
- bn
|
16 |
+
---
|
17 |
+
### Introduction
|
18 |
+
|
19 |
+
We introduce π MathOctopus, a series of open-source large language models (LLMs) specifically tailored for multilingual math problem-solving. The MathOctopus models are trained on π€ MGSM8KInstruct Dataset, encompassing ten distinct languages.
|
20 |
+
MathOctopus notably outperforms conventional open-source LLMs and exhibits superiority over ChatGPT in few-shot scenarios.
|
21 |
+
|
22 |
+
### Datasets
|
23 |
+
|
24 |
+
#### **MGSM8KInstruct**
|
25 |
+
|
26 |
+
| Training Dataset | En | Sw | Zh | Bn | De | Es | Fr | Ja | Ru | Th | Overall |
|
27 |
+
|:----------------------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|
|
28 |
+
| MGSM8KInstruct | 7473 | 7472 | 7466 | 6539 | 7466 | 7470 | 7469 | 7471 | 7361 | 7473 | **73.6K** |
|
29 |
+
|
30 |
+
|
31 |
+
#### **MSVAMP**
|
32 |
+
|
33 |
+
| Test Dataset | En | Sw | Zh | Bn | De | Es | Fr | Ja | Ru | Th | Overall |
|
34 |
+
|:----------------------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|
|
35 |
+
| MSVAMP | 1000 | 1000 | 1000 | 1000 | 1000 | 1000 | 1000 | 1000 | 1000 | 1000 | **10K** |
|
36 |
+
|
37 |
+
#### Usage
|
38 |
+
|
39 |
+
Our dataset and models are all available at Huggingface.
|
40 |
+
|
41 |
+
π€ [MGSM8KInstruct_Parallel Dataset](https://huggingface.co/datasets/Mathoctopus/GSM8KInstruct_Parallel)
|
42 |
+
|
43 |
+
π€ [MSVAMP Dataset](https://huggingface.co/datasets/Mathoctopus/MSVAMP)
|
44 |
+
|
45 |
+
|
46 |
+
## Models
|
47 |
+
|
48 |
+
| Base Model: LLama | Parallel-Training | Cross-Training |
|
49 |
+
|----|---------------------------------------------------------------|---------------------------------------------------------------------------|
|
50 |
+
| 7B-LLaMA 2 | π [MathOctopus-Parallel-7B](https://huggingface.co/Mathoctopus/Parallel_7B) | π [MathOctopus-Cross-7B](https://huggingface.co/Mathoctopus/Cross_7B) |
|
51 |
+
|| π[MathOctopus-Parallel-xRFT-7B](https://huggingface.co/Mathoctopus/Parallel_xRFT_7B)|π[MathOctopus-Cross-xRFT-7B](https://huggingface.co/Mathoctopus/Cross_xRFT_7B)|
|
52 |
+
| 13B-LLaMA 2 | π [MathOctopus-Parallel-13B] | π [MathOctopus-Cross-13B] |
|
53 |
+
|| π[MathOctopus-Parallel-xRFT-13B](https://huggingface.co/Mathoctopus/Parallel_xRFT_13B/tree/main)|π[MathOctopus-Cross-xRFT-13B]|
|
54 |
+
| 33B-LLaMA 1 | π [MathOctopus-Parallel-33B] | π [MathOctopus-Cross-33B] |
|
55 |
+
| 70B-LLaMA 2 | Coming soon! | Coming Soon! |
|
56 |
+
|
57 |
+
*-Parallel refers to our model trained with the parallel-training strategy.
|
58 |
+
|
59 |
+
*-Cross refers to our model trained with cross-training strategy.
|
60 |
+
|
61 |
+
*-xRFT means we train the model with multilingual rejection sampling.
|
62 |
+
### **Overall Results on MGSM**
|
63 |
+
| 7B Model | En | Sw | Zh | Bn | De | Es | Fr | Ja | Ru | Th | Overall |
|
64 |
+
|:--------------------------------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|
|
65 |
+
| MathOctopus<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 |
|
66 |
+
| **xRFT**-MathOctopus<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 |
|
67 |
+
| MathOctopus<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 |
|
68 |
+
| MathOctopus<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 |
|
69 |
+
| **xRFT**-MathOctopus<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 |
|
70 |
+
<p></p >
|
71 |
+
|
72 |
+
| 13B Model | En | Sw | Zh | Bn | De | Es | Fr | Ja | Ru | Th | Overall |
|
73 |
+
|:--------------------------------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|
|
74 |
+
| MathOctopus<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 |
|
75 |
+
| **xRFT**-MathOctopus<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 |
|
76 |
+
| MathOctopus<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 |
|
77 |
+
| **xRFT**-MathOctopus<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 |
|
78 |
+
<p></p >
|
79 |
+
|
80 |
+
| 30-34B Model | En | Sw | Zh | Bn | De | Es | Fr | Ja | Ru | Th | Overall |
|
81 |
+
|:--------------------------------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|
|
82 |
+
| MathOctopus<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 |
|
83 |
+
| **xRFT**-MathOctopus<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 |
|
84 |
+
| MathOctopus<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 |
|
85 |
+
| **xRFT**-MathOctopus<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 |
|
86 |
+
### **Overall Results on MSVAMP**
|
87 |
+
| 7B Model | En | Sw | Zh | Bn | De | Es | Fr | Ja | Ru | Th | Overall |
|
88 |
+
|:--------------------------------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|
|
89 |
+
| MathOctopus<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 |
|
90 |
+
| **xRFT**-MathOctopus<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 |
|
91 |
+
| MathOctopus<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 |
|
92 |
+
| MathOctopus<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 |
|
93 |
+
| **xRFT**-MathOctopus<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 |
|
94 |
+
<p></p >
|
95 |
+
|
96 |
+
| 13B Model | En | Sw | Zh | Bn | De | Es | Fr | Ja | Ru | Th | Overall |
|
97 |
+
|:--------------------------------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|
|
98 |
+
| MathOctopus<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 |
|
99 |
+
| **xRFT**-MathOctopus<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 |
|
100 |
+
| MathOctopus<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 |
|
101 |
+
| **xRFT**-MathOctopus<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 |
|
102 |
+
<p></p >
|
103 |
+
|
104 |
+
| 30-34B Model | En | Sw | Zh | Bn | De | Es | Fr | Ja | Ru | Th | Overall |
|
105 |
+
|:--------------------------------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|:--------|
|
106 |
+
| MathOctopus<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 |
|
107 |
+
| **xRFT**-MathOctopus<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 |
|
108 |
+
| MathOctopus<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 |
|
109 |
+
| **xRFT**-MathOctopus<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 |
|
110 |
+
### **MathOctopus in English**
|
111 |
+
| Models | GSM8K | SVAMP |
|
112 |
+
|:--------------------------------|:--------|:--------|
|
113 |
+
| LLaMA 2-7B | 42.4 | 38.3 |
|
114 |
+
| MathOctopus<sup>P</sup>-7B | 49.3 | 46.8 |
|
115 |
+
| MathOctopus<sup>C</sup>-7B | 50.8 | 49.3 |
|
116 |
+
| LLaMA 2-13B | 51.0 | 50.9 |
|
117 |
+
| MathOctopus<sup>P</sup>-13B | 55.5 | 52.1 |
|
118 |
+
| MathOctopus<sup>C</sup>-13B | 56.6 | 56.6 |
|
119 |
+
| LLaMA 1-33B | 50.0 | 49.0 |
|
120 |
+
| MathOctopus<sup>P</sup>-33B | 56.0 | 52.5 |
|
121 |
+
| MathOctopus<sup>C</sup>-33B | 53.7 | 51.5 |
|
122 |
+
## Intended Uses
|
123 |
+
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.
|