---
license: llama3
base_model:
  - meta-llama/Meta-Llama-3-8B
library_name: transformers
tags:
  - llama-3
  - merge
  - mergekit
  - logihertz
  - nyra
  - orchestrator
  - omni-core
  - independent-research
model_type: merge
pipeline_tag: text-generation
widget:
  - text: "Analyze the following Python code for logical errors, explain the bugs creatively, and output the corrected script in strict JSON format."
    example_title: "Omni-Disciplinary Task"
---

# 🌐 Nyra-Master: The Apex Orchestrator

**Nyra-Master** is the flagship unified model developed by **Logihertz Systems OPC Pvt Ltd**. As the pinnacle of the independent **Nyra Project**, this model serves as the "Apex Orchestrator." It seamlessly integrates the specialized capabilities of the entire Nyra suite: the rigid logic of **Tier A**, the expansive contextual creativity of **Tier B**, and the precise tool-execution of **Tier C**.

## 🛠 Model Specifications

* **Developer:** Logihertz Systems
* **Lead Architect:** Sameer Tawade
* **Project Status:** Independent Research
* **Architecture:** Optimized Llama-3-8B (Transformer-based Omni-Merge)
* **Merge Methodology:** Linear Merge (Optimized for multi-domain holistic reasoning)
* **Language(s):** English (Primary), Multi-language Code (Python, C++, JS, etc.)

## 🎯 Intended Use Cases

Nyra-Master is engineered for highly complex, multi-step workflows that require dynamic intent switching:

* **Universal Orchestration:** Acting as the primary router in multi-agent systems, dynamically shifting between creative, logical, and executable states.
* **Complex Pipeline Reasoning:** Handling prompts that require simultaneous math execution, creative explanation, and strict formatting.
* **General Purpose Excellence:** Serving as a standalone, highly capable assistant for developers, researchers, and enterprise environments.

## 📊 Evaluation & Benchmarking Matrix

*This flagship model is currently undergoing rigorous evaluation across all major AI domains. Scores are marked as pending while the self-verified evaluation pipeline completes.*

| **Category** | **Benchmark** | **Metric** | **Score** | **Status** |
| :--- | :--- | :--- | :--- | :--- |
| **Holistic Reasoning** | MMLU-Pro | 5-shot Accuracy | *Pending* | Eval in Progress |
| **Multi-Turn Chat** | MT-Bench | Average Score | *Pending* | Eval in Progress |
| **Code Execution** | HumanEval | Pass@1 | *Pending* | Eval in Progress |
| **Instruction Strictness**| IFEval | Prompt-level Strict | *Pending* | Eval in Progress |
| **Graduate Logic** | GPQA | 0-shot Accuracy | *Pending* | Eval in Progress |
| **Advanced Math** | MATH | 4-shot Chain-of-Thought| *Pending* | Eval in Progress |

## 💻 Implementation

To run Nyra-Master locally, ensure you have the latest `transformers` library installed. Due to its dense knowledge retention, we recommend running this model in `float16` precision.

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

model_id = "logihertz/nyra-Master"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalGeneration.from_pretrained(
    model_id, 
    torch_dtype=torch.float16, 
    device_map="auto"
)

prompt = "Explain quantum superposition. Then, write a Python script simulating a coin flip to represent the concept, and format the output as a JSON object."
inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
outputs = model.generate(**inputs, max_new_tokens=1024)

print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```

## ⚖️ Limitations & Ethical Considerations

Nyra-Master is released under the Llama 3 Community License. While designed to be an omni-capable orchestrator, combining logic, code, and creative capabilities can occasionally lead to complex hallucinations in highly ambiguous edge cases. Users should implement secondary validation systems for critical deployments.