---
base_model: unsloth/qwen2.5-coder-1.5b-instruct
tags:
- text-generation-inference
- transformers
- unsloth
- qwen2
- trl
license: apache-2.0
language:
- en
datasets:
- Daemontatox/math_conv
library_name: transformers
model-index:
- name: Zirel_1.5
  results:
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: IFEval (0-Shot)
      type: wis-k/instruction-following-eval
      split: train
      args:
        num_few_shot: 0
    metrics:
    - type: inst_level_strict_acc and prompt_level_strict_acc
      value: 41.68
      name: averaged accuracy
    source:
      url: https://huggingface.co./spaces/open-llm-leaderboard/open_llm_leaderboard#/?search=Daemontatox%2FZirel_1.5
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: BBH (3-Shot)
      type: SaylorTwift/bbh
      split: test
      args:
        num_few_shot: 3
    metrics:
    - type: acc_norm
      value: 15.08
      name: normalized accuracy
    source:
      url: https://huggingface.co./spaces/open-llm-leaderboard/open_llm_leaderboard#/?search=Daemontatox%2FZirel_1.5
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: MATH Lvl 5 (4-Shot)
      type: lighteval/MATH-Hard
      split: test
      args:
        num_few_shot: 4
    metrics:
    - type: exact_match
      value: 11.33
      name: exact match
    source:
      url: https://huggingface.co./spaces/open-llm-leaderboard/open_llm_leaderboard#/?search=Daemontatox%2FZirel_1.5
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: GPQA (0-shot)
      type: Idavidrein/gpqa
      split: train
      args:
        num_few_shot: 0
    metrics:
    - type: acc_norm
      value: 1.34
      name: acc_norm
    source:
      url: https://huggingface.co./spaces/open-llm-leaderboard/open_llm_leaderboard#/?search=Daemontatox%2FZirel_1.5
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: MuSR (0-shot)
      type: TAUR-Lab/MuSR
      args:
        num_few_shot: 0
    metrics:
    - type: acc_norm
      value: 3.33
      name: acc_norm
    source:
      url: https://huggingface.co./spaces/open-llm-leaderboard/open_llm_leaderboard#/?search=Daemontatox%2FZirel_1.5
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: MMLU-PRO (5-shot)
      type: TIGER-Lab/MMLU-Pro
      config: main
      split: test
      args:
        num_few_shot: 5
    metrics:
    - type: acc
      value: 12.71
      name: accuracy
    source:
      url: https://huggingface.co./spaces/open-llm-leaderboard/open_llm_leaderboard#/?search=Daemontatox%2FZirel_1.5
      name: Open LLM Leaderboard
---


# Zireal 1.5 - Fast Reasoning Model

**Developed by:** Daemontatox  
**Finetuned from:** [unsloth/qwen2.5-coder-1.5b-instruct](https://huggingface.co./unsloth/qwen2.5-coder-1.5b-instruct)  
**License:** Apache 2.0  

## Overview  
**Zireal 1.5** is a **fast, efficient reasoning model** designed for structured problem-solving and mathematical inference. It has been fine-tuned using **GRPO (General Reinforcement Policy Optimization)** on **24,000 high-quality mathematical examples**, making it highly effective for step-by-step reasoning and logic-based tasks.  

## Features  
- **Optimized for fast, structured reasoning** with minimal computational overhead.  
- **GRPO-trained** for superior decision-making in mathematical contexts.  
- **Lightweight yet highly capable**, leveraging Qwen2.5's instruction-tuned efficiency.  
- **Ideal for logic, algebra, arithmetic, and structured problem-solving.**  

## Usage  
You can load **Zireal 1.5** using `transformers`:  

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "Daemontatox/Zireal-1.5"  # Replace with actual model name
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name)

inputs = tokenizer("Solve: 3x - 7 = 11", return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=50)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```

## Intended Use  
- **Mathematical reasoning** (algebra, arithmetic, logic-based problems).  
- **Step-by-step structured problem-solving** for computational tasks.  
- **Lightweight inference** for fast, efficient reasoning applications.  

## Limitations  
- Primarily designed for structured reasoning rather than open-ended text generation.  
- Best suited for logic and mathematics rather than creative or conversational AI.  

## Acknowledgments  
**Zireal 1.5** is part of the **Zireal** model series, focusing on efficient and necessary reasoning. It is built on **Qwen2.5** and optimized using **Unsloth** for high-performance inference.  

---  

🔗 **[Hugging Face Model Card](https://huggingface.co./Daemontatox/Zireal-1.5)** (Replace with actual link)  
📜 **License:** Apache 2.0  
```
# [Open LLM Leaderboard Evaluation Results](https://huggingface.co./spaces/open-llm-leaderboard/open_llm_leaderboard)
Detailed results can be found [here](https://huggingface.co./datasets/open-llm-leaderboard/Daemontatox__Zirel_1.5-details)!
Summarized results can be found [here](https://huggingface.co./datasets/open-llm-leaderboard/contents/viewer/default/train?q=Daemontatox%2FZirel_1.5&sort[column]=Average%20%E2%AC%86%EF%B8%8F&sort[direction]=desc)!

|      Metric       |Value (%)|
|-------------------|--------:|
|**Average**        |    14.24|
|IFEval (0-Shot)    |    41.68|
|BBH (3-Shot)       |    15.08|
|MATH Lvl 5 (4-Shot)|    11.33|
|GPQA (0-shot)      |     1.34|
|MuSR (0-shot)      |     3.33|
|MMLU-PRO (5-shot)  |    12.71|