File size: 3,549 Bytes
ea4a33f
0ea2a9b
 
 
 
fd36686
 
 
 
 
 
0ea2a9b
fd36686
0ea2a9b
 
 
 
 
 
 
 
fd36686
0ea2a9b
ea4a33f
 
 
 
 
0ea2a9b
 
 
 
 
 
 
 
 
 
 
ea4a33f
0ea2a9b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8d0a1bc
0ea2a9b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
25c111c
 
 
 
 
0ea2a9b
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
125
126
127
128
129
---
base_model: Qwen/Qwen3-1.7B
library_name: transformers
model_name: Vex_Amber_mini_2.5
tags:
- generated_from_trainer
- trl
- sft
- code
- reasoning
- 2B
licence: license
license: cc-by-nc-4.0
language:
- en
- fa
- fr
metrics:
- code_eval
new_version: Arioron/Vex-Amber-Mini-1.2
pipeline_tag: text-generation
num_parameters: 2000000000
---





      type: text-generation
      name: Mathematical Reasoning
    dataset:
      name: MATH
      type: math
      split: test
    metrics:
    - name: Accuracy
      type: accuracy
      value: 55.0
---

# Amber Fable 1.0

## Model Description
**Amber Fable 1.0** is a **1.7B parameter** specialized language model, fine-tuned using **LoRA (Low-Rank Adaptation)** on the powerful **Qwen3-1.7B** base model. 

This model is engineered specifically for **mathematical reasoning** and **algorithmic logic**. It achieves remarkable performance on math benchmarks (75% on GSM8K) for its size class, making it a highly efficient solution for educational tools and logic-based tasks, although it trades off some general world knowledge (MMLU) to achieve this peak reasoning capability.

- **Developed by:** Arioron
- **Model type:** Decoder-only Transformer (LoRA Adapter)
- **Language(s):** English
- **License:** Apache 2.0
- **Finetuned from model:** Qwen/Qwen3-1.7B

### Model Sources
- **Repository:** https://huggingface.co/Arioron/Amber-Fable-1.0
- **Documentation:** Arioron Model Docs

## Performance

Amber Fable 1.0 demonstrates state-of-the-art efficiency in mathematical tasks.

| Benchmark | Metric | Score | Description |
| :--- | :--- | :--- | :--- |
| **GSM8K** | Accuracy | **75.0%** | Grade School Math |
| **MATH** | Accuracy | **55.0%** | Advanced Math Problems |
| **HumanEval**| Pass@1 | **42.0%** | Python Coding Capability |
| MMLU | Accuracy | 22.0% | General World Knowledge |



## Quick Start

```python
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

model_name = "Arioron/Amber-Fable-1.0"

tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype=torch.float16,
    device_map="auto"
)

# Math reasoning example
messages = [
    {"role": "user", "content": "Natalia sold clips to 48 of her friends in April, and then she sold half as many clips in May. How many clips did Natalia sell altogether in April and May?"},
]

input_text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(input_text, return_tensors="pt").to(model.device)

outputs = model.generate(
    **inputs,
    max_new_tokens=512,
    temperature=0.6,
    do_sample=True,
    top_p=0.9,
    pad_token_id=tokenizer.eos_token_id
)

print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```
### Model Summary
- **Model:** Amber Fable 1.0 (1.7B)
- **Specialty:** Advanced Math Reasoning
- **Logic:** Chain-of-Thought (CoT)
- **Coding:** Python & Algorithms (42%)
- **Tuning:** LoRA on Synthetic/Textbooks
- **Base:** Qwen3-1.7B (PyTorch/PEFT)
- **Usage:** Tutoring, Puzzles & Scripts
- **Caution:** Verify all calculations
- **Author:** Arioron (2025)
If you use this model in your research, please cite:
code
Bibtex
@misc{amberfable1.0,
  title = {Amber Fable 1.0: A Specialized 1.7B Math Model},
  author = {Arioron},
  year = {2025},
  publisher = {Hugging Face},
  howpublished = {\url{https://huggingface.co/Arioron/Amber-Fable-1.0}}

  contact
  Email: inquiry@arioron.com
Website: https://arioron.com
Documentation: https://docs.arioron.com
}