Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,43 @@
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
+
base_model: qwen2.5-math-1.5b
|
| 4 |
+
tags:
|
| 5 |
+
- text-generation
|
| 6 |
+
- mathematics
|
| 7 |
+
- physics
|
| 8 |
+
- morphy
|
| 9 |
+
- moolvylabs
|
| 10 |
+
language:
|
| 11 |
+
- en
|
| 12 |
---
|
| 13 |
+
|
| 14 |
+
**Morphy-Math-1.5B** is the third model in the Morphy lineup developed by the **Moolvylabs** team. Unlike previous text-focused versions, this model has undergone selective, highly targeted fine-tuning on complex advanced academic datasets. It is fully optimized for exceptional performance in higher sciences while remaining lightweight enough to run locally on consumer-grade hardware.
|
| 15 |
+
|
| 16 |
+
---
|
| 17 |
+
|
| 18 |
+
## ✨ Strengths (What the model excels at)
|
| 19 |
+
|
| 20 |
+
* 🧮 **Higher Mathematics & Calculus** — outstanding performance in solving university-level limits, improper integrals, derivatives, and infinite series step-by-step using strict Chain of Thought (CoT).
|
| 21 |
+
* 📐 **Linear Algebra & Differential Equations** — effortlessly handles non-linear differential equations (such as Bernoulli equations), spectral theorems, matrix eigenvalues, and eigenvectors.
|
| 22 |
+
* ⚛️ **Quantum Computing & Physics** — natively understands advanced physics, Dirac bra-ket notations, Pauli operators, and complex quantum state transformations.
|
| 23 |
+
|
| 24 |
+
---
|
| 25 |
+
|
| 26 |
+
## ⚠️ Important Features and Limitations (Disclaimer)
|
| 27 |
+
|
| 28 |
+
Since this model is deployed across various quantization formats to support low-end hardware, please keep the following nuances in mind:
|
| 29 |
+
|
| 30 |
+
* 🔬 **Model Quantization Impact:** The full **FP16** and **Q8_0** versions retain near 100% of the model's complex reasoning logic. However, aggressive bit-compression variants (**Q4_K_M** / **Q5_K_M**) may occasionally exhibit minor logical degradation or index confusion on highly abstract, multi-variable academic equations (e.g., advanced tensor calculus).
|
| 31 |
+
* 🌐 **Language Nuance:** For maximum accuracy in complex symbolic formulas and advanced scientific notations, prompting the model in English is highly recommended.
|
| 32 |
+
|
| 33 |
+
---
|
| 34 |
+
|
| 35 |
+
## 🛠️ Technical Specifications
|
| 36 |
+
|
| 37 |
+
| Parameter | Value |
|
| 38 |
+
| :--- | :--- |
|
| 39 |
+
| **Name** | Morphy-Math-1.5B |
|
| 40 |
+
| **Generation** | Third model in the family |
|
| 41 |
+
| **Base Model** | Qwen2.5-Math-1.5B-Instruct |
|
| 42 |
+
| **Data Type** | Custom Higher Mathematics & Quantum Physics Paks |
|
| 43 |
+
| **Purpose** | Advanced Calculus / Tensor Analysis / Symbolic Reasoning |
|