This model is used only for local interpretation (intent + slot extraction) inside the Vesper app. It does NOT perform financial calculations or direct data access.

🔧 Usage (Vesper)

The Vesper app will:

Fetch manifest.json
Compare versions
Download the model if needed
Verify checksum
Run locally via a GGUF runtime (e.g., llama.cpp)

This model is downloaded on-demand by users as the “Offline Assistant Pack”.

📦 Files

models/qwen2.5-0.5b-instruct-q4.gguf — quantized model
manifest.json — app metadata (version, checksum, size)

🧠 Scope

This model is used for:

intent detection
domain classification
slot extraction
turn classification (e.g., confirm, edit, follow-up)

It is not used for:

financial computations
database access
final response generation

⚠️ Notes

This is a mirrored distribution, not the original source.
Files are versioned and controlled for application stability.
Do not rely on this repo for general AI usage.

🔗 Original Model

Upstream repository:

https://huggingface.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF

All credit belongs to the original authors.

📜 License

This repository redistributes the model under its original license.

See LICENSE for full terms.

rhou-and-ry/vesper-qwen2.5-0.5b-q4

Vesper Assistant Pack — Qwen2.5 0.5B (Q4 GGUF)