persian-embeddings-gguf

Persian-English Text Embedding Model (GGUF)

This is a GGUF conversion of the original heydariAI/persian-embeddings from heydariAI/persian-embeddings. The GGUF version is optimized for efficient inference with frameworks like llama.cpp and integration with Ollama.

Model Details

• Original Model: heydariAI/persian-embeddings heydariAI/persian-embeddings
• Model Type: Text Embedding
• Languages: Persian (Farsi) and English
• Conversion: Converted to GGUF format for efficient CPU/GPU inference

Intended Uses

As the original repository mentions, this model is designed for:

• Semantic similarity search
• Retrieval-Augmented Generation (RAG) systems
• Text clustering and classification
• Feature extraction for Persian and English texts

Ollama Integration

This model is available on Ollama Hub as alishendi/persian-embeddings. Ollama Model Page Here's how to use it:

Installation and Usage

1. First, install Ollama:

curl -fsSL https://ollama.com/install.sh | sh

1. Pull the model:

ollama pull alishendi/persian-embeddings

1. Generate embeddings vector for a sample text:

curl http://localhost:11434/api/embeddings \
  -H "Content-Type: application/json" \
  -d '{
    "model": "alishendi/persian-embeddings",
    "prompt": "برای این متن، بردار بازنمایی تولید کن."
  }'

Python Example

Here's how to use the model with Ollama in Python:

# pip install ollama

import ollama
import numpy as np

# Generate embedding
response = ollama.embeddings(
    model="alishendi/persian-embeddings",
    prompt="برای این متن، بردار بازنمایی تولید کن."
)

embedding = np.array(response["embedding"])
print(f"Embedding shape: {embedding.shape}")
print(f"First 10 values: {embedding[:10]}")

Also you can download the model on LM Studio and generate the embeddings via its API

Performance

The GGUF format provides efficient inference on both CPU and GPU hardware. Performance will vary based on your hardware configuration.

Acknowledgements

This model is based on the original work from heydariAI/persian-embeddings. Please consider citing the original authors if you use this model in your research.