Update README.md

README.md

@@ -1,15 +1,13 @@
 ---
-base_model:
-- Daemontatox/SphinX
+base_model: unsloth/qwen2.5-7b-instruct-bnb-4bit
 tags:
 - text-generation-inference
 - transformers
 - unsloth
 - qwen2
 - trl
-- logic
-- Reasoning
 - COT
+- Reasoning
 license: apache-2.0
 language:
 - en
@@ -17,18 +15,10 @@ datasets:
 - Daemontatox/LongCOT-Reason
 metrics:
 - accuracy
-- recall
 - bleu
-- brier_score
-- code_eval
 - character
-- charcut_mt
-- cer
 - bleurt
-- chrf
-pipeline_tag: text-generation
 library_name: transformers
-new_version: Daemontatox/Sphinx2.0
 ---
 
 ![image](./image.webp)
@@ -42,82 +32,97 @@ new_version: Daemontatox/Sphinx2.0
 
 ## Model Overview  
 
-The **Super Strong Reasoning Model** is a high-performance AI designed for complex reasoning and decision-making tasks. It builds on the robust Qwen2.5 architecture, finetuned with cutting-edge methods to ensure exceptional capabilities in speed, accuracy, and logical reasoning.  
+The **Super Strong Reasoning Model** is an advanced AI system optimized for logical reasoning, multi-step problem-solving, and decision-making tasks. Designed with efficiency and accuracy in mind, it employs a structured system prompt to ensure high-quality answers through a transparent and iterative thought process.  
 
-### Key Features  
-- **Advanced Reasoning:** Specially trained for logical, abstract, and multi-step reasoning.  
-- **Speed Optimization:** Training accelerated 2x using [Unsloth](https://github.com/unslothai/unsloth), resulting in faster deployment cycles.  
-- **Precision Efficiency:** Utilizes bnb-4bit precision for low-resource environments without performance trade-offs.  
-- **Wide Applicability:** Performs well across a broad range of tasks, including natural language understanding, creative generation, and structured problem-solving.  
+### System Prompt and Workflow  
+
+This model operates using an innovative reasoning framework structured around the following steps:  
+
+1. **Initial Thought:**  
+   The model uses `<Thinking>` tags to reason step-by-step and craft its best possible response.  
+   Example:  
+
+2. **Self-Critique:**  
+It evaluates its initial response within `<Critique>` tags, focusing on:  
+- **Accuracy:** Is it factually correct and verifiable?  
+- **Clarity:** Is it clear and free of ambiguity?  
+- **Completeness:** Does it fully address the request?  
+- **Improvement:** What can be enhanced?  
+Example:  
+
+3. **Revision:**  
+Based on the critique, the model refines its response within `<Revising>` tags.  
+Example:  
 
-[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
+4. **Final Response:**  
+The revised response is presented clearly within `<Final>` tags.  
+Example:  
+
+5. **Tag Innovation:**  
+When needed, the model creates and defines new tags for better structuring or clarity, ensuring consistent usage.  
+Example:  
+
+### Key Features  
+- **Structured Reasoning:** Transparent, multi-step approach for generating and refining answers.  
+- **Self-Improvement:** Built-in critique and revision ensure continuous response enhancement.  
+- **Clarity and Adaptability:** Tagging system provides organized, adaptable responses tailored to user needs.  
+- **Creative Flexibility:** Supports dynamic problem-solving with the ability to introduce new tags and concepts.  
 
 ---
 
 ## Use Cases  
 
-This model can be employed in various domains:  
-1. **Research and Analysis:** Extract insights, synthesize data, and assist in knowledge discovery.  
-2. **Business Decision-Making:** Streamline complex decisions with AI-driven recommendations.  
-3. **Education and Tutoring:** Provide step-by-step explanations and reasoning for academic problems.  
-4. **Creative Writing and Content Generation:** Develop detailed, logical, and engaging content.  
-5. **Game Design and Puzzles:** Solve and create logical challenges, puzzles, or scenarios.  
+The model is designed for various domains, including:  
+1. **Research and Analysis:** Extracting insights and providing structured explanations.  
+2. **Education:** Assisting with tutoring by breaking down complex problems step-by-step.  
+3. **Problem-Solving:** Offering logical and actionable solutions for multi-step challenges.  
+4. **Content Generation:** Producing clear, well-organized creative or professional content.  
 
 ---
 
 ## Training Details  
 
-### Training Frameworks  
-- **Primary Tools:**  
-  - [Unsloth](https://github.com/unslothai/unsloth) for accelerated training.  
-  - Hugging Face Transformers and the TRL library for reinforcement learning with human feedback (RLHF).  
+- **Frameworks:**  
+- [Unsloth](https://github.com/unslothai/unsloth) for accelerated training.  
+- Hugging Face Transformers and the TRL library for reinforcement learning with human feedback (RLHF).  
 
-### Dataset and Preprocessing  
-The model was finetuned on a carefully curated dataset of reasoning-focused tasks, ensuring its ability to handle:  
-- Logical puzzles and mathematical problems.  
-- Complex question-answering tasks.  
-- Deductive and inductive reasoning scenarios.  
+- **Dataset:** Finetuned on diverse reasoning-focused tasks, including logical puzzles, mathematical problems, and commonsense reasoning scenarios.  
 
-### Hardware and Efficiency  
-- **Precision:** Trained with bnb-4bit quantization for memory efficiency.  
-- **Speed Gains:** Leveraged optimized kernels to achieve 2x faster training while maintaining robustness and high accuracy.  
+- **Hardware Efficiency:**  
+- Trained with bnb-4bit precision for reduced memory usage.  
+- Optimized training pipeline achieving 2x faster development cycles.  
 
 ---
 
-## Model Performance  
+## Performance Metrics  
 
-### Benchmarks  
-This model achieves superior results on key reasoning benchmarks:  
-- **ARC (AI2 Reasoning Challenge):** Outperforms baseline models by a significant margin.  
-- **GSM8K (Math Reasoning):** High accuracy in multi-step problem-solving.  
-- **CommonsenseQA:** Robust understanding of commonsense reasoning tasks.  
-
-### Metrics  
-- **Accuracy:** Consistently high on logical and abstract reasoning benchmarks.  
-- **Inference Speed:** Optimized for real-time applications.  
-- **Resource Efficiency:** Low memory footprint, suitable for deployment in limited-resource environments.  
+The model excels in reasoning benchmarks:  
+- **ARC (AI2 Reasoning Challenge):** High accuracy in logical and commonsense tasks.  
+- **GSM8K (Math Reasoning):** Superior results in multi-step problem-solving.  
+- **CommonsenseQA:** Strong comprehension of everyday reasoning tasks.  
 
 ---
 
 ## Ethical Considerations  
 
-While this model is highly capable, its deployment should align with ethical guidelines:  
-1. **Transparency:** Ensure users understand its reasoning limitations.  
-2. **Bias Mitigation:** While trained on diverse data, outputs should be evaluated for fairness.  
-3. **Safe Usage:** Avoid applications that may harm individuals or propagate misinformation.  
+- **Transparency:** Responses are structured for verifiability through tagging.  
+- **Bias Mitigation:** Includes self-critique to minimize biases and ensure fairness.  
+- **Safe Deployment:** Users are encouraged to evaluate outputs to prevent harm or misinformation.  
 
 ---
 
 ## License  
 
-This model is open-source and distributed under the Apache 2.0 license. Users are encouraged to adapt and share the model, provided they comply with the license terms.  
+This model is distributed under the Apache 2.0 license, allowing users to use, modify, and share it in compliance with the license terms.  
+
+---
 
 ## Acknowledgments  
 
 Special thanks to:  
-- [Unsloth](https://github.com/unslothai/unsloth) for enabling accelerated training workflows.  
-- Hugging Face for providing the foundational tools and libraries.  
+- [Unsloth](https://github.com/unslothai/unsloth) for accelerated training workflows.  
+- Hugging Face for their powerful tools and libraries.  
 
 ---
 
-Experience the power of reasoning like never before. Leverage the **Super Strong Reasoning Model** for your AI-driven solutions today!
+Experience the **Super Strong Reasoning Model**, leveraging its structured reasoning and self-improvement capabilities for any task requiring advanced AI reasoning.