Have a ComfyUI workflow? Get a readable Workflow Doc in your browser.Make a workflow readable.Try it now ↗
Quantized builds of decibel-hq/reso1-3b-en (merged LoRA) for llama.cpp / LM Studio / local inference.
Model source
Source excerpt
Quantized builds of decibel-hq/reso1-3b-en (merged LoRA) for llama.cpp / LM Studio / local inference.
Sources
1 sourceVerified Aug 15
Model artifacts
1 artifactSource excerpts
2 excerpts--- license: apache-2.0 language: - en base_model: decibel-hq/reso1-3b-en pipeline_tag: text-to-speech tags: [text-to-speech, tts, orpheus, snac, gguf, llama.cpp] --- # Reso-1 3B English — GGUF Quantized builds of [decibel-hq/reso1-3b-en](https://huggingface.co/decibel-hq/reso1-3b-en) (merged LoRA) for llama.cpp / LM Studio / local inference. | file | quant | size | use | |---|---|---|---| | reso1-q4_k_m.gguf | Q4_K_M (5.05 bpw) | 2.09 GB | recommended; near-realtime on Apple Silicon | Note: this generates **audio tokens**, not text — decode via SNAC 24 kHz (7 tokens/frame from id 128266). Reference implementation: [`scripts/say_local.py`](https://github.com/decibel-hq/reso/blob/main/scripts/say_local.py). Prompt `reso: <text>`, tags `<laugh> <sigh> ...`, `repeat_penalty >= 1.1`. Research preview — private; not licensed for redistribution.
Source context: 21 downloads · 0 likes · Pipeline text-to-speech · Repo decibel-hq/reso1-3b-en-GGUF
