Update README.md

README.md

@@ -1,96 +1,77 @@
-# Model Card: PathfinderAI 32b
-
-## Model Overview
-PathfinderAI 32b is a state-of-the-art large language model fine-tuned for advanced reasoning tasks. It is optimized for chain-of-thought (CoT) reasoning and excels in solving complex, multi-step problems. The model has demonstrated superior performance on the MATH dataset, achieving second place on Hugging Face, outperforming nearly 2,000 other models.
+---
+base_model:
+- Qwen/QwQ-32B-Preview
+tags:
+- text-generation-inference
+- transformers
+- unsloth
+- qwen2
+- trl
+license: apache-2.0
+language:
+- en
+metrics:
+- accuracy
+new_version: Daemontatox/CogitoZ
+library_name: transformers
+---
 
-- **Model Size**: 32 billion parameters
-- **Architecture**: Transformer-based
-- **Base Model**: Qwen-0.5B
-- **Fine-Tuning Dataset**: MATH and other reasoning-intensive corpora
-- **Performance**: 
-  - Ranked #2 on Hugging Face for the MATH dataset
-  - Real-time text-based reasoning capabilities
+# CogitoZ - Qwen2
 
----
-![image](./image.webp)
-## Intended Use
-### Primary Use Cases
-- **Education**: Provide detailed, step-by-step solutions to mathematical and logical problems.
-- **Research**: Assist with hypothesis generation, logical inference, and data-driven research.
-- **Business**: Support complex decision-making processes, financial modeling, and strategic planning.
-- **Legal and Policy Analysis**: Simplify legal texts, analyze policy documents, and construct logical arguments.
-- **Healthcare**: Aid in diagnostic reasoning and structured decision support.
+## Model Overview
 
-### Out-of-Scope Use Cases
-- Generating harmful, biased, or unethical content.
-- Applications involving real-time physical world interactions (e.g., robotics or autonomous systems).
+CogitoZ - Qwen2 is a state-of-the-art large language model fine-tuned to excel in advanced reasoning and real-time decision-making tasks. This enhanced version was trained using [Unsloth](https://github.com/unslothai/unsloth), achieving a 2x faster training process. Leveraging Hugging Face's TRL (Transformers Reinforcement Learning) library, CogitoZ combines efficiency with exceptional reasoning performance.
 
----
+- **Developed by**: Daemontatox
+- **License**: Apache 2.0
+- **Base Model**: [Qwen/QwQ-32B-Preview](https://huggingface.co/Qwen/QwQ-32B-Preview)
+- **Finetuned from**: [Daemontatox/CogitoZ](https://huggingface.co/Daemontatox/CogitoZ)
 
-## Model Details
-- **Model Type**: Fine-tuned Transformer
-- **Training Objective**: Optimized for reasoning and logical problem solving using chain-of-thought prompts.
-- **Language Support**: Multilingual, with an emphasis on English reasoning tasks.
-- **Quantization**: Supports 8-bit and 4-bit quantization for deployment on resource-constrained devices (e.g., Raspberry Pi 5).
-- **Key Features**:
-  - Advanced chain-of-thought (CoT) reasoning
-  - Enhanced multi-step problem-solving ability
-  - Scalable deployment on edge devices
+[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
 
 ---
 
-## Training Data
-The model was fine-tuned on:
-- **MATH Dataset**: A large-scale dataset of mathematical problems with diverse complexity levels.
-- **Additional Corpora**: Custom curated datasets focusing on logical reasoning, structured decision-making, and multi-domain inference.
+## Key Features
 
-### Data Limitations
-The training data may contain biases inherent to the datasets, which could impact the model's predictions in specific contexts.
+1. **Fast Training**: Optimized with Unsloth, achieving a 2x faster training cycle without compromising model quality.
+2. **Enhanced Reasoning**: Utilizes advanced chain-of-thought (CoT) reasoning for solving complex problems.
+3. **Quantization Ready**: Supports 8-bit and 4-bit quantization for deployment on resource-constrained devices.
+4. **Scalable Inference**: Seamless integration with text-generation-inference tools for real-time applications.
 
 ---
 
-## Performance Metrics
-### Benchmark Results
-- **MATH Dataset**: Achieved a near-perfect accuracy of X% (specific score) on challenging problem sets.
-- **Hugging Face Leaderboard**: Secured second place among nearly 2,000 models.
+## Intended Use
 
-### Strengths
-- Exceptional logical inference and reasoning.
-- Robust performance in multi-step, complex tasks.
+### Primary Use Cases
+- **Education**: Real-time assistance for complex problem-solving, especially in mathematics and logic.
+- **Business**: Supports decision-making, financial modeling, and operational strategy.
+- **Healthcare**: Enhances diagnostic accuracy and supports structured clinical reasoning.
+- **Legal Analysis**: Simplifies complex legal documents and constructs logical arguments.
 
 ### Limitations
-- May struggle with tasks outside its reasoning-focused training scope.
-- Potential biases from training data may influence specific outputs.
+- May produce biased outputs if the input prompts contain prejudicial or harmful content.
+- Should not be used for real-time, high-stakes autonomous decisions (e.g., robotics or autonomous vehicles).
 
 ---
 
-## Deployment and Usage
-### Inference
-PathfinderAI 32b can be deployed using:
-- **Hugging Face Transformers Library**:
-  ```python
-  from transformers import AutoModelForCausalLM, AutoTokenizer
-
-  model_name = "PathfinderAI/32b-reasoning-model"
-  tokenizer = AutoTokenizer.from_pretrained(model_name)
-  model = AutoModelForCausalLM.from_pretrained(model_name)
+## Technical Details
 
-  prompt = "Solve the equation: x^2 - 5x + 6 = 0. Step-by-step:"
-  inputs = tokenizer(prompt, return_tensors="pt")
-  outputs = model.generate(**inputs)
-  print(tokenizer.decode(outputs[0], skip_special_tokens=True))
-```
+- **Training Framework**: Hugging Face's Transformers and TRL libraries.
+- **Optimization Framework**: Unsloth for faster and efficient training.
+- **Language Support**: English.
+- **Quantization**: Compatible with 8-bit and 4-bit inference modes for deployment on edge devices.
 
-Edge Devices: Optimized for quantized inference (8-bit, 4-bit) on resource-constrained devices like Raspberry Pi 5.
-Ethical Considerations
-Bias Mitigation: While efforts have been made to reduce biases during fine-tuning, users should be cautious of potential unintended outputs.
-Safe Usage: The model should not be used for applications promoting harm, misinformation, or unethical practices.
+### Deployment Example
 
+#### Using Hugging Face Transformers:
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
 
+model_name = "Daemontatox/CogitoZ"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(model_name)
 
-@article{PathfinderAI2025,
-  title={PathfinderAI 32b: A State-of-the-Art Reasoning Model},
-  author={Ammar Alnagar and contributors},
-  journal={Hugging Face Leaderboard},
-  year={2025}
-}
+prompt = "Explain the Pythagorean theorem step-by-step:"
+inputs = tokenizer(prompt, return_tensors="pt")
+outputs = model.generate(**inputs)
+print(tokenizer.decode(outputs[0], skip_special_tokens=True))