Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
pipeline_tag: text-generation
|
| 3 |
+
tags:
|
| 4 |
+
- NPU
|
| 5 |
+
---
|
| 6 |
+
# Granite-4.0-Micro
|
| 7 |
+
Run **Granite-4.0-Micro** optimized for **Qualcomm NPUs** with [nexaSDK](https://sdk.nexa.ai) on Android.
|
| 8 |
+
|
| 9 |
+
## Model Description
|
| 10 |
+
**Granite-4.0-Micro** is a 3B parameter instruction-tuned model in the Granite 4.0 family, developed by IBM.
|
| 11 |
+
It’s optimized for **long-context reasoning (128K tokens)**, efficient inference, and enterprise-ready capabilities such as tool calling and retrieval-augmented generation. The model balances compact size with strong performance across general NLP tasks, making it suitable for both experimentation and production workloads.
|
| 12 |
+
|
| 13 |
+
## Features
|
| 14 |
+
- **Compact transformer architecture**: 3B parameters with GQA, RoPE, SwiGLU, and RMSNorm layers.
|
| 15 |
+
- **Instruction-following & tool calling**: Tuned with supervised finetuning, alignment (RLHF), and model merging for robust enterprise tasks.
|
| 16 |
+
- **Multilingual support**: Covers 12+ languages including English, German, Spanish, French, Japanese, Korean, Arabic, and Chinese.
|
| 17 |
+
- **Extended context window**: Supports sequences up to **128K tokens** for long-form reasoning.
|
| 18 |
+
|
| 19 |
+
## Use Cases
|
| 20 |
+
- Conversational AI and virtual assistants.
|
| 21 |
+
- Enterprise applications needing **tool/API calling** and structured outputs.
|
| 22 |
+
- Long-document summarization, classification, and extraction.
|
| 23 |
+
- Retrieval-augmented generation (RAG) for knowledge-intensive workflows.
|
| 24 |
+
- Lightweight coding assistants and multilingual dialog systems.
|
| 25 |
+
|
| 26 |
+
## Inputs and Outputs
|
| 27 |
+
**Input**: Natural language text prompts, chat conversations, or tool-augmented requests.
|
| 28 |
+
**Output**: Natural language responses—answers, explanations, summaries, structured JSON for function calls, or code snippets.
|
| 29 |
+
|
| 30 |
+
## License
|
| 31 |
+
This model is released under the **Creative Commons Attribution–NonCommercial 4.0 (CC BY-NC 4.0)** license.
|
| 32 |
+
Non-commercial use, modification, and redistribution are permitted with attribution.
|
| 33 |
+
For commercial licensing, please contact **dev@nexa.ai**.
|