OraQwen3-Omni-20B-A3B-Instruct-FP8

OraQwen3-Omni-20B-A3B-Instruct-FP8

This repository contains the OraQwen3-Omni-20B-A3B-Instruct-FP8 model developed by Ora Computing. This is the FP8-quantized version of oracomputing/OraQwen3-Omni-20B-A3B-Instruct, which is itself a compressed and fine-tuned derivative of the original Qwen/Qwen3-Omni-30B-A3B-Instruct model.

---

Model Overview

Model name: OraQwen3-Omni-20B-A3B-Instruct-FP8
Source model (immediate parent): oracomputing/OraQwen3-Omni-20B-A3B-Instruct (this model is the FP8 quantization of it)
Original upstream model: Qwen/Qwen3-Omni-30B-A3B-Instruct
Derived size: ~20 billion total parameters / ~3 billion active parameters (MoE; compressed from the upstream model's ~30 billion total)
Quantization: FP8 (W8A8 — 8-bit weights and activations) quantization, performed with llm-compressor, for reduced memory footprint and faster inference
Purpose: Evaluation/test-use only; optimized for internal benchmarking and non-production integration.
License: See LICENSE (Custom Model License Agreement)

---

Intended Use & Restrictions

Permitted use

• Internal testing, benchmarking, and evaluation of the model by the named Licensee.
• Exploration of model behaviours, prompt engineering, and non-production prototypes.
• Evaluation across supported modalities: text, audio, image, and video input; text and audio output.

Prohibited use

• Deployment in a production or commercial service, publicly-facing API, resale, or redistribution.
• Fine-tuning or creating derivative models for production use without a separate agreement.
• Disclosure or sharing of the model (or its weights) to third parties beyond the named Licensee.

Out-of-scope use

• Use of the model in regulated or safety-critical contexts (unless separately permitted).
• Any use that violates the Apache License, Version 2.0 under which the upstream model is distributed.

---

Compression, Fine-Tuning & Quantization

• Parameter count compression: Total parameter count reduced from 30 billion to 20 billion by compressing MLP layers; the MoE active-parameter count (3B) is unchanged.
• Fine-tuning: The compressed model was subsequently fine-tuned on proprietary data curated by Ora Computing.
• FP8 quantization: The fine-tuned model was quantized to FP8 (W8A8 — 8-bit weights and activations) using llm-compressor to reduce the memory footprint and accelerate inference.

---

Limitations & Risks

• Compressed and quantized models may not replicate the full behaviour of the base model under all prompt categories, particularly domain-specific or rare inputs.
• The model is provided as-is for testing only and is not certified for production use.
• Users should validate outputs carefully and monitor for bias or unintended behaviours.

---

Upstream Attribution

This model is derived from the Qwen3-Omni-30B-A3B-Instruct model released by Alibaba Cloud under the Apache License, Version 2.0.

"Copyright 2025 Alibaba Cloud. Licensed under the Apache License, Version 2.0."

For full terms, see: https://github.com/QwenLM/Qwen3-Omni/blob/main/LICENSE
Apache License, Version 2.0: https://www.apache.org/licenses/LICENSE-2.0

---

Contact & Support

For licensing inquiries or to request extended evaluation rights, please contact:
[email protected]

---

Repository and model access are regulated. Do not redistribute or share without explicit written permission from Ora Computing.