File size: 2,517 Bytes
e375964
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
license: mit
library_name: transformers
---
# Titan-Math-Pro
<!-- markdownlint-disable first-line-h1 -->
<!-- markdownlint-disable html -->
<!-- markdownlint-disable no-duplicate-header -->

<div align="center">
  <img src="figures/fig1.png" width="60%" alt="Titan-Math-Pro" />
</div>
<hr>

<div align="center" style="line-height: 1;">
  <a href="LICENSE" style="margin: 2px;">
    <img alt="License" src="figures/fig2.png" style="display: inline-block; vertical-align: middle;"/>
  </a>
</div>

## 1. Introduction

Titan-Math-Pro is a mathematics-first model trained on verified proofs and competition problems with process supervision.

## 2. Evaluation Results

### Comprehensive Benchmark Results

<div align="center">

| | Benchmark | NumeriX | Titan-base | ProofNet | Titan-Math-Pro |
|---|---|---|---|---|---|
| **Core Reasoning Tasks** | Math Reasoning | 0.589 | 0.595 | 0.577 | 0.633 |
|  | Logical Reasoning | 0.816 | 0.818 | 0.793 | 0.850 |
|  | Common Sense | 0.749 | 0.735 | 0.776 | 0.783 |
| **Language Understanding** | Reading Comprehension | 0.707 | 0.731 | 0.743 | 0.760 |
|  | Question Answering | 0.595 | 0.595 | 0.642 | 0.649 |
|  | Text Classification | 0.815 | 0.817 | 0.831 | 0.849 |
|  | Sentiment Analysis | 0.779 | 0.780 | 0.765 | 0.818 |
| **Generation Tasks** | Code Generation | 0.703 | 0.682 | 0.702 | 0.730 |
|  | Creative Writing | 0.693 | 0.670 | 0.658 | 0.700 |
|  | Dialogue Generation | 0.683 | 0.677 | 0.656 | 0.702 |
|  | Summarization | 0.793 | 0.765 | 0.771 | 0.805 |
| **Specialized Capabilities** | Translation | 0.802 | 0.815 | 0.814 | 0.826 |
|  | Knowledge Retrieval | 0.667 | 0.672 | 0.687 | 0.718 |
|  | Instruction Following | 0.756 | 0.750 | 0.741 | 0.798 |
|  | Safety Evaluation | 0.762 | 0.739 | 0.734 | 0.779 |

</div>

### Overall Performance Summary
The Titan-Math-Pro demonstrates strong performance across all evaluated benchmark categories, with particularly notable results in reasoning and generation tasks.

## 3. Chat Website & API Platform
We offer a chat interface and API for you to interact with Titan-Math-Pro. Please check our official website for more details.

## 4. How to Run Locally

Please refer to our code repository for more information about running Titan-Math-Pro locally.

### Temperature
We recommend setting the temperature parameter to 0.6.

## 5. License
This repository is released under the mit license. The model supports commercial use.

## 6. Contact
If you have any questions, please contact us at math@titan-compute.com.