Dirdx/YuE2-3B-MLX-with-Hum-encoder

text-to-audio model by Dirdx

Open original source ↗

From the publisher

Model card & documentation

Source preview

Read the publisher’s intended use, setup instructions, evaluations and limitations. The original model card is the source of truth.

Native MLX port of m-a-p/YuE2-3B (AR/NAR Mixture-of-Transformers music generation) with the YuE2-Vae decoder folded in. This Repo contains an additional Vae Encoder model, which is used for the Hum lora feature in my pinokio app: https://github.com/Rdx-ai-art/yue2-mlx.pinokio. Runs on Apple Silicon with no PyTorch dependency. Lyrics + style prompt in, 48 kHz stereo song out, with optional editable ABC score planning. Upstream repo : https://huggingface.co/ahmadw/YuE2-3B-MLX The NAR (flow-matching) path stays at 8 bits in the 4-bit variant because fully 4-bit flow matching drifts (12% velocity error). VAE decode is bit-exact in all variants. 8bit/ is the recommended default: near-bf16 output at roughly twice the decode speed. Each folder contains model.safetensors, vae.safetensors, config.json, vaeconfig.json, qwen.tiktoken and yue2generationconfig.json. The Python files at the repo root are the inference code. Options: --cot off|melody|full (symbolic ABC planning; full is the upstream default), --abc-file to supply your own score, --cfg-scale, --steps (NAR midpoint steps, default 32),…

Dirdx/YuE2-3B-MLX-with-Hum-encoder on Hugging Face A short preview, not the full document.

Read the full model card ↗ · Preview checked 2026-09-23T07:17:24.341Z

Inside the original model card — Document outline
  1. YuE2-3B MLX-with-Hum-Encoder
  2. Variants
  3. Usage
  4. Converting yourself
  5. License

Headings are captured from the source. Links open the publisher’s document, not a locally hosted copy.

Documentation belongs to its respective authors. Reported project/model license: cc-by-nc-4.0. A listing is not a grant of reuse or training rights. Confirm the document’s own terms at the source.

Model overview

Dirdx/YuE2-3B-MLX-with-Hum-encoder is a text-to-audio model repository published by Dirdx on Hugging Face. The source reports the mlx library.

This page summarizes Hub metadata. For intended use, training data, evaluation results and limitations, consult the original model card.

Model facts

Task
text-to-audio
Library
mlx
Recent downloads (30 days)
0
Cumulative likes
0
Hugging Face trending score
0
Reported safetensors parameters
Not reported
Architecture
Not reported
License
cc-by-nc-4.0
Access
Not gated by Hugging Face
Created
2026-09-23T06:33:09.000Z
Last modified
2026-09-23T06:33:10.000Z

Compatibility and lineage

Reported languages: en, zh

Reported base models: m-a-p/YuE2-3B

Parameter count is not a RAM/VRAM requirement. Check precision, quantization, context length and runtime compatibility in the model card; no hardware or API-cost claim is inferred here.

Use and evaluate

Check license terms and access requirements first. Review the model card’s documented loading instructions, evaluations and safety limitations. Benchmark scores are not imported by this directory, and popularity is not an accuracy ranking.

No model weights are downloaded or executed by AltAPIs. Never enable remote model code without reviewing it.

Model card and usage instructions

Model files and configuration

Community discussion

Source and freshness

Source: Hugging Face Hub. Metadata observed 2026-09-23T06:37:25.545Z. Daily imports are snapshots, not real-time monitoring.

Popularity and source listings do not establish security, suitability, licensing rights or benchmark performance.

Open original source

Related models

audio-cpp/Yue2-3B-GGUF

m-a-p/YuE2-3B