README.md

---
license: apache-2.0
language:
- en
base_model:
- mistralai/Devstral-Small-2-24B-Instruct-2512
pipeline_tag: text-generation
library_name: transformers
tags:
- reasoning
- coding
- math
- science
- instruction-tuned
- mistral
- pytorch
---

# **Supertron2-24B: A Capable Instruction-Tuned Coding and Reasoning Model**

## **Model Description**

**Supertron2-24B** is an instruction-tuned language model built on top of [mistralai/Devstral-Small-2-24B-Instruct-2512](https://huggingface.co/mistralai/Devstral-Small-2-24B-Instruct-2512). It is designed for practical coding assistance, structured reasoning, math, science, general chat, and everyday instruction following.

* **Developed by:** Surpem
* **Model type:** Causal Language Model
* **Architecture:** Dense Transformer, 24B parameters
* **License:** Apache 2.0

---

## **Capabilities**

### **Coding**

Supertron2-24B is designed to help write, explain, and debug code. It can assist with practical programming tasks, implementation planning, error analysis, and code review style explanations.

### **Reasoning**

The model can work through multi-step questions, compare options, follow structured instructions, and produce concise answers when requested.

### **Math**

Supertron2-24B can handle arithmetic, algebra-style problems, word problems, and step-by-step mathematical explanations.

### **Science**

The model can explain scientific concepts clearly, answer STEM questions, and help with educational or technical writing.

### **General Chat**

Supertron2-24B can assist with writing, brainstorming, explanations, planning, summarization, and general everyday questions.

---


## **Get Started**

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

model_id = "Surpem/Supertron2-24B"

tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForImageTextToText.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto",
    trust_remote_code=True,
)

messages = [
    {"role": "user", "content": "Write a Python function that checks if a string is a palindrome."}
]

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 | 48 GB | 80 GB+ |
| 4-bit quantized | 16 GB | 24 GB+ |

For long contexts or larger batches, use more VRAM or reduce batch size and max sequence length.

---

## **Intended Use**

Supertron2-24B is intended for:

* Coding assistance
* Software engineering reasoning
* Math and science explanations
* General chat and instruction following
* Writing, summarization, and brainstorming
* Research and technical assistance

---

## **Limitations**

* The model can make mistakes and should be checked for important work.
* It may produce incorrect code, incomplete reasoning, or outdated information.
* It should not be used as the only source for medical, legal, financial, or safety-critical decisions.
* Generated code should be reviewed and tested before use.

---

## **Citation**

```bibtex
@misc{surpem2026supertron2-24b,
      title={Supertron2-24B -- Instruction-Tuned Coding and Reasoning Model},
      author={Surpem},
      year={2026},
      url={https://huggingface.co/Surpem/Supertron2-24B},
}
```