Hugging Face's logo

enosislabs
/
math-mini-1.7b-preview-16bits

Text Generation
Transformers
PyTorch
Safetensors
English
qwen3
text-generation-inference
math
enosis-labs
math-mini
vllm
conversational
Model card Files
xet
Community
1
math-mini-1.7b-preview-16bits
File size: 4,304 Bytes 
1a5be19
61f1682
 
 
 
 
 
 
 
 
 
 
1a5be19
 
 
61f1682
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
 
---
base_model: Qwen/Qwen3-1.7B
model_name: Math Mini 1.7B (Preview)
tags:
- text-generation
- text-generation-inference
- transformers
- qwen3
- math
- enosis-labs
- math-mini
- vllm # Tag for vLLM version
license: apache-2.0
language:
- en
---

# Math Mini 1.7B (Preview)

**Math Mini 1.7B (Preview)** is a larger, more capable model developed by **Enosis Labs** as part of the "Mini Series." Building on the foundation of the 0.6B version, this 1.7B model delivers significantly improved performance, deeper reasoning, and greater accuracy in mathematical tasks. It is fine-tuned from the original `Qwen/Qwen3-1.7B` base model (not from Unsloth's pre-adapted versions).

## Philosophy & Capabilities

The Mini Series, along with the "Enosis Math" and "Enosis Code" models, incorporates step-by-step reasoning by default, enabling more efficient, clear, and well-founded answers. All models in the Math series have been trained with carefully curated step-by-step problem-solving datasets, resulting in a greater ability to reason and explain solutions in a structured way.

**Math Mini 1.7B (Preview)** is optimized for:

* **Basic and Intermediate Algebra:** Solving equations, manipulating expressions, and handling more complex algebraic problems.
* **Arithmetic & Sequential Reasoning:** Calculations and breaking down problems into logical steps, with improved multi-step reasoning.
* **Elementary & Intermediate Logic:** Applying deduction in mathematical contexts, now with broader coverage.
* **Competition Problem Solving (Introductory to Intermediate):** Enhanced foundational and competition-style skills, adapted to the increased model scale.

Larger models in the "Enosis Math" series address even more advanced topics such as calculus, higher algebra, and olympiad problems. The "Code Mini" and "Enosis Code" series are oriented towards programming and algorithmic tasks, maintaining the same philosophy of explicit and efficient reasoning.

This model is a **preview version** and is under continuous improvement and evaluation.

## Quick Start

Available in Hugging Face Transformers format and for high-throughput inference servers like vLLM.

### vLLM (Inference Server)

Install vLLM:

```bash
pip install vllm
```

Start the vLLM server with the model (16-bit version):

```bash
vllm serve "enosislabs/math-mini-1.7b-preview-16bits"
```

Call the server using curl:

```bash
curl -X POST "http://localhost:8000/v1/chat/completions" \
    -H "Content-Type: application/json" \
    --data '{
        "model": "enosislabs/math-mini-1.7b-preview-16bits",
        "messages": [
            {"role": "user", "content": "What is the capital of France?"}
        ]
    }'
```

### Transformers (Hugging Face)

Use a pipeline as a high-level helper:

```python
from transformers import pipeline

pipe = pipeline("text-generation", model="enosislabs/math-mini-1.7b-preview-16bits")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)
```

## Prompt Format (Qwen3 ChatML)

For best results, use the Qwen3 ChatML format. The `tokenizer.apply_chat_template` method handles this automatically.

```text
<|im_start|>system
You are a helpful AI assistant. Provide a detailed step-by-step solution.
<|im_end|>
<|im_start|>user
{user_question}
<|im_end|>
<|im_start|>assistant
```

## Acknowledgements

* Fine-tuned from the original `Qwen/Qwen3-1.7B` base model.
* Training process accelerated and optimized using [Unsloth](https://github.com/unslothai/unsloth) for efficiency, but not using Unsloth's pre-adapted weights.

## Citation

If you use this model, please cite:

```bibtex
@software{enosislabs_math_mini_1.7b_preview_2025,
  author = {{Enosis Labs}},
  title = {{Math Mini 1.7B (Preview)}},
  year = {2025},
  publisher = {Hugging Face},
  version = {0.1-preview},
  url = {https://huggingface.co/enosislabs/math-mini-1.7b-preview-16bits}
}
```

<!--
Key points:
- Now 1.7B, with improved performance and reasoning over 0.6B.
- Fine-tuned from the original Qwen3-1.7B, not Unsloth's pre-adapted weights.
- Emphasizes default activation of step-by-step reasoning across the series.
- Clear and modern examples for vLLM and Transformers.
- ChatML prompt is central to the experience.
- Assumes the repo contains the 1.7B model for both vLLM and Transformers.
-->