97 lines
3.5 KiB
Markdown
97 lines
3.5 KiB
Markdown
---
|
|
license: apache-2.0
|
|
language:
|
|
- en
|
|
base_model:
|
|
- Qwen/Qwen3-1.7B
|
|
pipeline_tag: text-generation
|
|
library_name: transformers
|
|
tags:
|
|
- reasoning
|
|
- math
|
|
- coding
|
|
- instruction-tuned
|
|
- pytorch
|
|
---
|
|
# **Supertron2-1.7B: A Compact, Efficient Instruction-Tuned Language Model**
|
|
## **Model Description**
|
|
**Supertron2-1.7B** is an instruction-tuned language model built on top of Qwen3-1.7B. Designed to be a **reliable, efficient daily driver**, it delivers strong performance across math, coding, reasoning, science, general knowledge, and general conversation while remaining lightweight enough to run on consumer hardware.
|
|
|
|
* **Developed by:** Surpem
|
|
* **Model type:** Causal Language Model
|
|
* **Architecture:** Dense Transformer, 1.7B parameters
|
|
* **Fine-tuned from:** [Qwen/Qwen3-1.7B](https://huggingface.co/Qwen/Qwen3-1.7B)
|
|
* **License:** Apache 2.0
|
|
|
|
---
|
|
|
|
## **Capabilities**
|
|
|
|
### **Reasoning**
|
|
Supertron2-1.7B is designed for clear multi-step reasoning, making it capable of breaking down complex problems in a structured and useful way. It can work through questions methodically rather than jumping directly to a final answer.
|
|
|
|
### **Math**
|
|
The model handles a range of math tasks, from arithmetic and algebra to word problems and structured problem solving. It is useful for explaining steps, checking calculations, and producing concise final answers.
|
|
|
|
### **Coding**
|
|
Supertron2-1.7B can write, debug, and explain code across popular languages including Python, JavaScript, C++, and more. It understands syntax, common programming patterns, algorithmic reasoning, and practical implementation details.
|
|
|
|
### **Science & General Knowledge**
|
|
Broad instruction tuning across science, STEM, and general knowledge domains means the model can hold technical conversations, explain difficult concepts clearly, and assist with research, writing, and analysis tasks.
|
|
|
|
### **Instruction Following**
|
|
The model is responsive to natural language instructions. Whether you need concise answers, detailed explanations, structured output, or creative writing, Supertron2-1.7B adapts to the format and tone you ask for without needing complex prompting tricks.
|
|
|
|
---
|
|
|
|
## **Get Started**
|
|
```python
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM
|
|
import torch
|
|
|
|
model_id = "Surpem/Supertron2-1.7B"
|
|
|
|
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
|
model = AutoModelForCausalLM.from_pretrained(
|
|
model_id,
|
|
torch_dtype=torch.bfloat16,
|
|
device_map="auto"
|
|
)
|
|
|
|
messages = [
|
|
{"role": "user", "content": "Explain the difference between LoRA and full fine-tuning."}
|
|
]
|
|
|
|
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
|
|
inputs = tokenizer(text, return_tensors="pt").to(model.device)
|
|
outputs = model.generate(**inputs, max_new_tokens=512)
|
|
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:], skip_special_tokens=True))
|
|
```
|
|
|
|
---
|
|
|
|
## **Hardware Requirements**
|
|
| Precision | Min VRAM | Recommended |
|
|
|---|---|---|
|
|
| bfloat16 | 5 GB | 8 GB+ |
|
|
| 4-bit quantized | 3 GB | 4 GB+ |
|
|
|
|
For 4-bit quantized inference:
|
|
```python
|
|
from transformers import BitsAndBytesConfig
|
|
|
|
bnb_config = BitsAndBytesConfig(load_in_4bit=True, bnb_4bit_compute_dtype=torch.bfloat16)
|
|
model = AutoModelForCausalLM.from_pretrained(model_id, quantization_config=bnb_config, device_map="auto")
|
|
```
|
|
|
|
---
|
|
|
|
## **Citation**
|
|
```bibtex
|
|
@misc{surpem2026supertron2-1.7b,
|
|
title={Supertron2-1.7B — Efficient Instruction-Tuned Language Model},
|
|
author={Surpem},
|
|
year={2026},
|
|
url={https://huggingface.co/Surpem/Supertron2-1.7B},
|
|
}
|
|
``` |