docs: document v1 quantized model revisions

README.md

@@ -20,16 +20,18 @@ This repository provides the pinned FireRedAudio model snapshot used by both:
 - [FireRedAudio T8star-Aix Desktop](https://github.com/T8mars/FireRedAudio-T8-Desktop)
 - [comfyui-fireredaudio-T8](https://github.com/T8mars/comfyui-fireredaudio-T8)
 
-The model files retain the upstream directory layout and are not converted or quantized. The source project and model are maintained by [FireRedTeam/FireRedAudio](https://github.com/FireRedTeam/FireRedAudio). This mirror exists to give the T8star-Aix integrations a stable, resumable download location. Review the upstream license, model documentation, and limitations before use.
+The source project and model are maintained by [FireRedTeam/FireRedAudio](https://github.com/FireRedTeam/FireRedAudio). This mirror provides pinned, resumable T8star-Aix revisions. Every revision keeps the same top-level layout and includes its own integrity and quantization metadata. Review the upstream license, model documentation, and limitations before use.
 
 ## Profiles
 
-| Profile | Download | Capabilities |
-|---|---:|---|
-| Lite | about 19.8 GiB | ASR and audio understanding |
-| Full | about 27.6 GiB | Lite plus TTS, voice design, and speech editing |
+| Revision | Download | Status | Capabilities |
+|---|---:|---|---|
+| `bf16-slim-v1` | about 21.19 GiB | stable, lossless | Full stack, original BF16 main weights and inference-only RedAE decoder |
+| `int8-wo-safe-v1` | about 14.76 GiB | stable, recommended | Full stack, TorchAO A16W8 Qwen blocks; audio modules stay in released precision |
+| `int8-wo-extended-v1` | about 13.58 GiB | experimental | Full stack, extended A16W8 including audio blocks |
+| `int8-convrot-experimental-v1` | about 14.76 GiB | experimental | Full stack, Comfy-Kitchen ConvRot A8W8 |
 
-Full adds `RedAE_decoder/model.pt`. Voice generation, design, and editing cannot run with Lite alone.
+All revisions above include `RedAE_decoder/model.safetensors`. GGUF is not provided because existing GGUF runtimes cannot execute the complete audio encoder, continuous latent, DiT, and RedAE graph.
 
 ## Required layout
 
@@ -39,17 +41,14 @@ The same directory can be used by Desktop and ComfyUI:
 FireRedAudio/
 ├─ FireRedAudio/
 │  ├─ config.json
-│  ├─ model-00001-of-00005.safetensors
-│  ├─ model-00002-of-00005.safetensors
-│  ├─ model-00003-of-00005.safetensors
-│  ├─ model-00004-of-00005.safetensors
-│  ├─ model-00005-of-00005.safetensors
+│  ├─ model-00001-of-*.safetensors
+│  ├─ ...
 │  ├─ model.safetensors.index.json
 │  ├─ processor_config.json
 │  ├─ tokenizer.json
 │  └─ tokenizer_config.json
 └─ RedAE_decoder/
-   └─ model.pt
+   └─ model.safetensors
 ```
 
 ## Desktop installation
@@ -63,8 +62,8 @@ from huggingface_hub import snapshot_download
 
 snapshot_download(
     repo_id="t8star/Firered-Audio-Comfy",
+    revision="int8-wo-safe-v1",
     local_dir=r"D:\FireRedAudio-Models",
-    allow_patterns=["FireRedAudio/*", "RedAE_decoder/*"],
 )
 ```
 
@@ -87,8 +86,8 @@ from huggingface_hub import snapshot_download
 
 snapshot_download(
     repo_id="t8star/Firered-Audio-Comfy",
+    revision="int8-wo-safe-v1",
     local_dir=r"D:\ComfyUI\models\TTS\FireRedAudio",
-    allow_patterns=["FireRedAudio/*", "RedAE_decoder/*"],
 )
 ```
 
@@ -96,7 +95,7 @@ Restart ComfyUI and select `FireRedAudio` in the `T8 FireRedAudio Model Loader`
 
 ## Integrity and reproducibility
 
-The integrations pin a specific immutable Hub revision. Expected sizes and SHA-256 values are stored in `manifests/model_firered_audio.json` in both GitHub repositories. The downloader performs a fast structural/size validation after download and can optionally perform a full SHA-256 validation.
+The integrations pin a specific immutable Hub revision. Every profile contains `fireredaudio-model.json` with its exact file sizes and SHA-256 values; INT8 profiles also contain `FireRedAudio/fireredaudio_quantization.json`. Desktop and ComfyUI perform a fast structural/size validation after selection and can optionally perform a full SHA-256 validation.
 
 ## Responsible use