81 lines
2.3 KiB
Markdown
81 lines
2.3 KiB
Markdown
|
---
|
|||
|
|
license: apache-2.0
|
|||
|
|
language:
|
|||
|
|
- en
|
|||
|
|
base_model:
|
|||
|
|
- Qwen/Qwen3-4B
|
|||
|
|
pipeline_tag: text-generation
|
|||
|
|
library_name: transformers
|
|||
|
|
tags:
|
|||
|
|
- reasoning
|
|||
|
|
- math
|
|||
|
|
- coding
|
|||
|
|
- instruction-tuned
|
|||
|
|
- pytorch
|
|||
|
|
---
|
|||
|
|
|
|||
|
|
# **Supertron1-4B: A Capable, Efficient Instruction-Tuned Language Model**
|
|||
|
|
|
|||
|
|
## **Model Description**
|
|||
|
|
|
|||
|
|
**Supertron1-4B** is an instruction-tuned language model built on top of Qwen3-4B. Designed to be a **reliable, efficient daily driver**, it delivers strong performance across math, coding, reasoning, and general conversation while remaining fast and lightweight enough to run on consumer hardware.
|
|||
|
|
|
|||
|
|
* **Developed by:** Surpem
|
|||
|
|
* **Model type:** Causal Language Model
|
|||
|
|
* **Architecture:** Dense Transformer, 4B parameters
|
|||
|
|
* **Fine-tuned from:** [Qwen/Qwen3-4B](https://huggingface.co/Qwen/Qwen3-4B)
|
|||
|
|
* **License:** Apache 2.0
|
|||
|
|
|
|||
|
|
---
|
|||
|
|
|
|||
|
|
## **Results**
|
|||
|
|
|
|||
|
|
Supertron1-4B holds its own against models in the 4–8B class and surpasses Mistral 7B on all four core benchmarks despite having nearly half the parameters.
|
|||
|
|
|
|||
|
|
<div align="center">
|
|||
|
|
<p align="center"><a href="https://postimg.cc/zLfkBN3D"><img width=800 src="https://i.postimg.cc/0NYXV2sS/1000037258.png"/></a></p>
|
|||
|
|
</div>
|
|||
|
|
|
|||
|
|
**Key takeaways:**
|
|||
|
|
- Beats Mistral 7B on every benchmark at 4B parameters
|
|||
|
|
- Strong GSM8K and HumanEval performance from math and coding focused tuning
|
|||
|
|
- Competitive with Phi-4 mini on a fraction of the compute
|
|||
|
|
|
|||
|
|
---
|
|||
|
|
|
|||
|
|
## **Get Started**
|
|||
|
|
|
|||
|
|
```python
|
|||
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM
|
|||
|
|
import torch
|
|||
|
|
|
|||
|
|
model_id = "surpem/supertron1-4b"
|
|||
|
|
|
|||
|
|
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))
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
## **Citation**
|
|||
|
|
|
|||
|
|
```bibtex
|
|||
|
|
@misc{surpem2026supertron1,
|
|||
|
|
title={Supertron1-4B — Efficient Instruction-Tuned Language Model},
|
|||
|
|
author={Surpem},
|
|||
|
|
year={2026},
|
|||
|
|
url={https://huggingface.co/surpem/supertron1-4b},
|
|||
|
|
}
|
|||
|
|
```
|