File size: 3,218 Bytes
f919a26 74074fe f919a26 74074fe f919a26 e219328 74074fe 054f375 f919a26 74074fe 054f375 74074fe 7fe1c9e 3566a7a 7fe1c9e 054f375 04f447d 054f375 e219328 74074fe e219328 054f375 3566a7a 054f375 e219328 054f375 e219328 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 | ---
library_name: gguf
license: apache-2.0
base_model: LiquidAI/LFM2.5-2.6B
tags:
- mew
- uncensored
- roleplay
- lfm
- linear-rnn
---
<div align="center">
<div style="display: flex; justify-content: center; gap: 1.5em; margin-bottom: 1.5em;">
<a href="https://huggingface.co/OrbitMC"><strong>👤 Profile</strong></a> •
<a href="https://huggingface.co/OrbitMC/Mew1-2.6B/tree/main/ggufs"><strong>📂 All Models</strong></a>
</div>
</div>
# 🐱 Mew1-2.6B
**Mew1-2.6B** is a high-performance, next-generation language model developed by **Unmid**. Engineered for elite intelligence and uncensored freedom, Mew provides unparalleled reasoning capabilities and creative depth at a compact 2.6B parameter scale.
Designed to push the limits of small-scale language modeling, Mew delivers instantaneous responses while maintaining a sophisticated grasp of complex instructions and nuanced roleplay. It excels in environments where low latency and high creative output are required.
## 🌟 Key Features
- **Uncensored by Design:** Optimized for unrestricted creative expression and objective reasoning.
- **Linear RNN Efficiency:** Built on a specialized architecture that offers linear scaling for long contexts.
- **Elite Reasoning:** Outperforms many models twice its size in logic and instruction following.
## 📦 GGUF Downloads
| File | Method | Estimated Size | Download | Best For |
| :--- | :--- | :---: | :---: | :--- |
| **Mew1-2.6B-q2-k.gguf** | Q2_K | 0.85 GB | [download](https://huggingface.co/OrbitMC/Mew1-2.6B/resolve/main/ggufs/Mew1-2.6B-q2-k.gguf?download=true) | Ultra-low RAM |
| **Mew1-2.6B-q4-k-m.gguf** | Q4_K_M | 1.56 GB | [download](https://huggingface.co/OrbitMC/Mew1-2.6B/resolve/main/ggufs/Mew1-2.6B-q4-k-m.gguf?download=true) | Balanced (Recommended) |
| **Mew1-2.6B-q6-k.gguf** | Q6_K | 2.15 GB | [download](https://huggingface.co/OrbitMC/Mew1-2.6B/resolve/main/ggufs/Mew1-2.6B-q6-k.gguf?download=true) | High Precision |
| **Mew1-2.6B-q8-0.gguf** | Q8_0 | 2.76 GB | [download](https://huggingface.co/OrbitMC/Mew1-2.6B/resolve/main/ggufs/Mew1-2.6B-q8-0.gguf?download=true) | Maximum Quality |
## 🛠️ Technical Specifications
- **Parameters:** 2.6 Billion
- **Architecture:** Fine-tuned Liquid Foundation Model (LFM) 2.5
- **Context Window:** 32,768 Tokens
- **Training Focus:** Advanced reasoning, creative writing, and technical instruction following.
## 🎯 Use Cases
- **Creative Writing:** Drafting stories, scripts, and roleplay scenarios with deep character consistency.
- **Technical Assistance:** Explaining complex concepts and generating code snippets.
- **Local Deployment:** Ideal for mobile devices, low-power laptops, and private server environments.
## ⚙️ Hardware Recommendations
| Quantization | Minimum RAM | Recommended Hardware |
| :--- | :---: | :--- |
| Q2_K | 2 GB | Most Mobile Phones |
| Q4_K_M | 4 GB | Standard Laptop / iPhone 13+ |
| Q8_0 | 6 GB | Dedicated GPU / 8GB+ Desktop RAM |
## 🚀 Usage Instructions
Compatible with **llama.cpp**, **LM Studio**, and **KoboldCPP**.
**Recommended Settings:**
- **Temperature:** 0.7 - 0.85 (for creativity)
- **Repeat Penalty:** 1.1
- **Top-P:** 0.9
---
*Developed with ❤️ by Unmid* |