README.md

---
license: gpl-3.0
language:
- en
datasets:
- Mxode/Magpie-Pro-10K-GPT4o-mini
pipeline_tag: text2text-generation
tags:
- chemistry
- biology
- finance
- legal
- music
- code
- climate
- medical
- text-generation-inference
---
# NanoLM-0.3B-Instruct-v2


English | [简体中文](README_zh-CN.md)


## Introduction

In order to explore the potential of small models, I have attempted to build a series of them, which are available in the [NanoLM Collections](https://huggingface.co/collections/Mxode/nanolm-66d6d75b4a69536bca2705b2).

This is NanoLM-0.3B-Instruct-v2. The model currently supports **English only**.



## Model Details

| Nano LMs | Non-emb Params | Arch | Layers | Dim | Heads | Seq Len |
| :----------: | :------------------: | :---: | :----: | :-------: | :---: | :---: |
| 25M         | 15M  |   MistralForCausalLM     | 12      | 312     | 12    |2K|
| 70M         | 42M |  LlamaForCausalLM          | 12     | 576    | 9   |2K|
| **0.3B**        | **180M** |  **Qwen2ForCausalLM**  | **12**   | **896**    | **14** | **4K** |
| 1B     | 840M | Qwen2ForCausalLM | 18   | 1536   | 12   |4K|

The tokenizer and model architecture of NanoLM-0.3B-Instruct-v1.1 are the same as [Qwen/Qwen2-0.5B](https://huggingface.co/Qwen/Qwen2-0.5B), but the number of layers has been reduced from 24 to 12.

As a result, NanoLM-0.3B-Instruct-v1.1 has only 0.3 billion parameters, with approximately **180 million non-embedding parameters**.

Despite this, NanoLM-0.3B-Instruct-v1.1 still demonstrates strong instruction-following capabilities.



## How to use

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

model_path = 'Mxode/NanoLM-0.3B-Instruct-v2'

model = AutoModelForCausalLM.from_pretrained(model_path).to('cuda:0', torch.bfloat16)
tokenizer = AutoTokenizer.from_pretrained(model_path)


def get_response(prompt: str, **kwargs):
    generation_args = dict(
        max_new_tokens = kwargs.pop("max_new_tokens", 512),
        do_sample = kwargs.pop("do_sample", True),
        temperature = kwargs.pop("temperature", 0.7),
        top_p = kwargs.pop("top_p", 0.8),
        top_k = kwargs.pop("top_k", 40),
        **kwargs
    )

    messages = [
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": prompt}
    ]
    text = tokenizer.apply_chat_template(
        messages,
        tokenize=False,
        add_generation_prompt=True
    )
    model_inputs = tokenizer([text], return_tensors="pt").to(model.device)

    generated_ids = model.generate(model_inputs.input_ids, **generation_args)
    generated_ids = [
        output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
    ]

    response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
    return response


prompt1 = "Calculate (4 - 1) * 7"
print(get_response(prompt1, do_sample=False))

"""
To calculate the expression (4 - 1) * 7, we need to follow the order of operations (PEMDAS):

1. Evaluate the expression inside the parentheses: 4 - 1 = 3
2. Multiply 3 by 7: 3 * 7 = 21

So, (4 - 1) * 7 = 21.
"""
```