Commit
•
1e93d42
1
Parent(s):
377f0a2
uploaded readme
Browse files
README.md
ADDED
@@ -0,0 +1,165 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
Quantization made by Richard Erkhov.
|
2 |
+
|
3 |
+
[Github](https://github.com/RichardErkhov)
|
4 |
+
|
5 |
+
[Discord](https://discord.gg/pvy7H8DZMG)
|
6 |
+
|
7 |
+
[Request more models](https://github.com/RichardErkhov/quant_request)
|
8 |
+
|
9 |
+
|
10 |
+
OpenMath-Mistral-7B-v0.1-hf - GGUF
|
11 |
+
- Model creator: https://huggingface.co/nvidia/
|
12 |
+
- Original model: https://huggingface.co/nvidia/OpenMath-Mistral-7B-v0.1-hf/
|
13 |
+
|
14 |
+
|
15 |
+
| Name | Quant method | Size |
|
16 |
+
| ---- | ---- | ---- |
|
17 |
+
| [OpenMath-Mistral-7B-v0.1-hf.Q2_K.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.Q2_K.gguf) | Q2_K | 2.53GB |
|
18 |
+
| [OpenMath-Mistral-7B-v0.1-hf.IQ3_XS.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.IQ3_XS.gguf) | IQ3_XS | 2.81GB |
|
19 |
+
| [OpenMath-Mistral-7B-v0.1-hf.IQ3_S.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.IQ3_S.gguf) | IQ3_S | 2.96GB |
|
20 |
+
| [OpenMath-Mistral-7B-v0.1-hf.Q3_K_S.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.Q3_K_S.gguf) | Q3_K_S | 2.95GB |
|
21 |
+
| [OpenMath-Mistral-7B-v0.1-hf.IQ3_M.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.IQ3_M.gguf) | IQ3_M | 3.06GB |
|
22 |
+
| [OpenMath-Mistral-7B-v0.1-hf.Q3_K.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.Q3_K.gguf) | Q3_K | 3.28GB |
|
23 |
+
| [OpenMath-Mistral-7B-v0.1-hf.Q3_K_M.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.Q3_K_M.gguf) | Q3_K_M | 3.28GB |
|
24 |
+
| [OpenMath-Mistral-7B-v0.1-hf.Q3_K_L.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.Q3_K_L.gguf) | Q3_K_L | 3.56GB |
|
25 |
+
| [OpenMath-Mistral-7B-v0.1-hf.IQ4_XS.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.IQ4_XS.gguf) | IQ4_XS | 3.67GB |
|
26 |
+
| [OpenMath-Mistral-7B-v0.1-hf.Q4_0.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.Q4_0.gguf) | Q4_0 | 3.83GB |
|
27 |
+
| [OpenMath-Mistral-7B-v0.1-hf.IQ4_NL.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.IQ4_NL.gguf) | IQ4_NL | 3.87GB |
|
28 |
+
| [OpenMath-Mistral-7B-v0.1-hf.Q4_K_S.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.Q4_K_S.gguf) | Q4_K_S | 3.86GB |
|
29 |
+
| [OpenMath-Mistral-7B-v0.1-hf.Q4_K.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.Q4_K.gguf) | Q4_K | 4.07GB |
|
30 |
+
| [OpenMath-Mistral-7B-v0.1-hf.Q4_K_M.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.Q4_K_M.gguf) | Q4_K_M | 4.07GB |
|
31 |
+
| [OpenMath-Mistral-7B-v0.1-hf.Q4_1.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.Q4_1.gguf) | Q4_1 | 4.24GB |
|
32 |
+
| [OpenMath-Mistral-7B-v0.1-hf.Q5_0.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.Q5_0.gguf) | Q5_0 | 4.65GB |
|
33 |
+
| [OpenMath-Mistral-7B-v0.1-hf.Q5_K_S.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.Q5_K_S.gguf) | Q5_K_S | 4.65GB |
|
34 |
+
| [OpenMath-Mistral-7B-v0.1-hf.Q5_K.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.Q5_K.gguf) | Q5_K | 4.78GB |
|
35 |
+
| [OpenMath-Mistral-7B-v0.1-hf.Q5_K_M.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.Q5_K_M.gguf) | Q5_K_M | 4.78GB |
|
36 |
+
| [OpenMath-Mistral-7B-v0.1-hf.Q5_1.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.Q5_1.gguf) | Q5_1 | 5.07GB |
|
37 |
+
| [OpenMath-Mistral-7B-v0.1-hf.Q6_K.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.Q6_K.gguf) | Q6_K | 5.53GB |
|
38 |
+
| [OpenMath-Mistral-7B-v0.1-hf.Q8_0.gguf](https://huggingface.co/RichardErkhov/nvidia_-_OpenMath-Mistral-7B-v0.1-hf-gguf/blob/main/OpenMath-Mistral-7B-v0.1-hf.Q8_0.gguf) | Q8_0 | 7.17GB |
|
39 |
+
|
40 |
+
|
41 |
+
|
42 |
+
|
43 |
+
Original model description:
|
44 |
+
---
|
45 |
+
license: apache-2.0
|
46 |
+
base_model:
|
47 |
+
- mistralai/Mistral-7B-v0.1
|
48 |
+
datasets:
|
49 |
+
- nvidia/OpenMathInstruct-1
|
50 |
+
language:
|
51 |
+
- en
|
52 |
+
tags:
|
53 |
+
- nvidia
|
54 |
+
- code
|
55 |
+
- math
|
56 |
+
---
|
57 |
+
|
58 |
+
|
59 |
+
# OpenMath-Mistral-7B-v0.1-hf
|
60 |
+
|
61 |
+
OpenMath models were designed to solve mathematical problems by integrating text-based reasoning with code blocks
|
62 |
+
executed by Python interpreter. The models were trained on [OpenMathInstruct-1](https://huggingface.co/datasets/nvidia/OpenMathInstruct-1),
|
63 |
+
a math instruction tuning dataset with 1.8M problem-solution pairs generated using permissively licensed
|
64 |
+
[Mixtral-8x7B](https://huggingface.co/mistralai/Mixtral-8x7B-v0.1) model.
|
65 |
+
|
66 |
+
<table border="1">
|
67 |
+
<tr>
|
68 |
+
<td></td>
|
69 |
+
<td colspan="2" style="text-align: center;">greedy</td>
|
70 |
+
<td colspan="2" style="text-align: center;">majority@50</td>
|
71 |
+
</tr>
|
72 |
+
<tr>
|
73 |
+
<td style="text-align: center;">model</td>
|
74 |
+
<td style="text-align: center;">GSM8K</td>
|
75 |
+
<td style="text-align: center;">MATH</td>
|
76 |
+
<td style="text-align: center;">GMS8K</td>
|
77 |
+
<td style="text-align: center;">MATH</td>
|
78 |
+
</tr>
|
79 |
+
<tr>
|
80 |
+
<td style="text-align: right;">OpenMath-CodeLlama-7B (<a href="https://huggingface.co/nvidia/OpenMath-CodeLlama-7b-Python">nemo</a> | <a href="https://huggingface.co/nvidia/OpenMath-CodeLlama-7b-Python-hf">HF</a>)</td>
|
81 |
+
<td style="text-align: center;">75.9</td>
|
82 |
+
<td style="text-align: center;">43.6</td>
|
83 |
+
<td style="text-align: center;">84.8</td>
|
84 |
+
<td style="text-align: center;">55.6</td>
|
85 |
+
</tr>
|
86 |
+
<tr>
|
87 |
+
<td style="text-align: right;">OpenMath-Mistral-7B (<a href="https://huggingface.co/nvidia/OpenMath-Mistral-7B-v0.1">nemo</a> | <a href="https://huggingface.co/nvidia/OpenMath-Mistral-7B-v0.1-hf">HF</a>)</td>
|
88 |
+
<td style="text-align: center;">80.2</td>
|
89 |
+
<td style="text-align: center;">44.5</td>
|
90 |
+
<td style="text-align: center;">86.9</td>
|
91 |
+
<td style="text-align: center;">57.2</td>
|
92 |
+
</tr>
|
93 |
+
<tr>
|
94 |
+
<td style="text-align: right;">OpenMath-CodeLlama-13B (<a href="https://huggingface.co/nvidia/OpenMath-CodeLlama-13b-Python">nemo</a> | <a href="https://huggingface.co/nvidia/OpenMath-CodeLlama-13b-Python-hf">HF</a>)</td>
|
95 |
+
<td style="text-align: center;">78.8</td>
|
96 |
+
<td style="text-align: center;">45.5</td>
|
97 |
+
<td style="text-align: center;">86.8</td>
|
98 |
+
<td style="text-align: center;">57.6</td>
|
99 |
+
</tr>
|
100 |
+
<tr>
|
101 |
+
<td style="text-align: right;">OpenMath-CodeLlama-34B (<a href="https://huggingface.co/nvidia/OpenMath-CodeLlama-34b-Python">nemo</a> | <a href="https://huggingface.co/nvidia/OpenMath-CodeLlama-34b-Python-hf">HF</a>)</td>
|
102 |
+
<td style="text-align: center;">80.7</td>
|
103 |
+
<td style="text-align: center;">48.3</td>
|
104 |
+
<td style="text-align: center;">88.0</td>
|
105 |
+
<td style="text-align: center;">60.2</td>
|
106 |
+
</tr>
|
107 |
+
<tr>
|
108 |
+
<td style="text-align: right;">OpenMath-Llama2-70B (<a href="https://huggingface.co/nvidia/OpenMath-Llama-2-70b">nemo</a> | <a href="https://huggingface.co/nvidia/OpenMath-Llama-2-70b-hf">HF</a>)</td>
|
109 |
+
<td style="text-align: center;"><b>84.7</b></td>
|
110 |
+
<td style="text-align: center;">46.3</td>
|
111 |
+
<td style="text-align: center;">90.1</td>
|
112 |
+
<td style="text-align: center;">58.3</td>
|
113 |
+
</tr>
|
114 |
+
<tr>
|
115 |
+
<td style="text-align: right;">OpenMath-CodeLlama-70B (<a href="https://huggingface.co/nvidia/OpenMath-CodeLlama-70b-Python">nemo</a> | <a href="https://huggingface.co/nvidia/OpenMath-CodeLlama-70b-Python-hf">HF</a>)</td>
|
116 |
+
<td style="text-align: center;">84.6</td>
|
117 |
+
<td style="text-align: center;"><b>50.7</b></td>
|
118 |
+
<td style="text-align: center;"><b>90.8</b></td>
|
119 |
+
<td style="text-align: center;"><b>60.4</b></td>
|
120 |
+
</tr>
|
121 |
+
</table>
|
122 |
+
|
123 |
+
The pipeline we used to produce these models is fully open-sourced!
|
124 |
+
|
125 |
+
- [Code](https://github.com/Kipok/NeMo-Skills)
|
126 |
+
- [Models](https://huggingface.co/collections/nvidia/openmath-65c5619de2ba059be0775014)
|
127 |
+
- [Dataset](https://huggingface.co/datasets/nvidia/OpenMathInstruct-1)
|
128 |
+
|
129 |
+
See our [paper](https://arxiv.org/abs/2402.10176) for more details!
|
130 |
+
|
131 |
+
# How to use the models?
|
132 |
+
|
133 |
+
Try to [run inference with our models](https://github.com/Kipok/NeMo-Skills/blob/main/docs/inference.md) with just a few commands!
|
134 |
+
|
135 |
+
# Reproducing our results
|
136 |
+
|
137 |
+
We provide [all instructions](https://github.com/Kipok/NeMo-Skills/blob/main/docs/reproducing-results.md) to fully reproduce our results.
|
138 |
+
|
139 |
+
# Improving other models
|
140 |
+
|
141 |
+
To improve other models or to learn more about our code, read through the docs below.
|
142 |
+
|
143 |
+
- [NeMo-Skills Pipeline](https://github.com/Kipok/NeMo-Skills)
|
144 |
+
- [Generating synthetic data](https://github.com/Kipok/NeMo-Skills/blob/main/docs/synthetic-data-generation.md)
|
145 |
+
- [Finetuning models](https://github.com/Kipok/NeMo-Skills/blob/main/docs/finetuning.md)
|
146 |
+
- [Evaluating models](https://github.com/Kipok/NeMo-Skills/blob/main/docs/evaluation.md)
|
147 |
+
|
148 |
+
In our pipeline we use [NVIDIA NeMo](https://www.nvidia.com/en-us/ai-data-science/generative-ai/nemo-framework/),
|
149 |
+
an end-to-end, cloud-native framework to build, customize, and deploy generative AI models anywhere.
|
150 |
+
It includes training and inferencing frameworks, guardrailing toolkits, data curation tools, and pretrained models,
|
151 |
+
offering enterprises an easy, cost-effective, and fast way to adopt generative AI.
|
152 |
+
|
153 |
+
# Citation
|
154 |
+
|
155 |
+
If you find our work useful, please consider citing us!
|
156 |
+
|
157 |
+
```bibtex
|
158 |
+
@article{toshniwal2024openmath,
|
159 |
+
title = {OpenMathInstruct-1: A 1.8 Million Math Instruction Tuning Dataset},
|
160 |
+
author = {Shubham Toshniwal and Ivan Moshkov and Sean Narenthiran and Daria Gitman and Fei Jia and Igor Gitman},
|
161 |
+
year = {2024},
|
162 |
+
journal = {arXiv preprint arXiv: Arxiv-2402.10176}
|
163 |
+
}
|
164 |
+
```
|
165 |
+
|