Upload folder using huggingface_hub

README.md

@@ -37,11 +37,11 @@ just fetch-tiny-tts && cargo run -p rlx-tiny-tts --release --features apple-sili
 
 ## Contents
 
-Either use `tiny-tts.rlxp` (official RLXPFLAT asset pack) or the ONNX + `frontend/` tree. Also used by `rlx-melotts`.
+Either use `tiny-tts.rlxp` (native nested graph packs + frontend) or a local ONNX pack source. Also used by `rlx-melotts`. Runtime does **not** load `.onnx` from Hub.
 
 ## Pack layout (`.rlxp`)
 
-Asset-only pack (no weight tensors in the TOC): VITS2 ONNX subgraphs plus English frontend (G2P, BERT tokenizer, CMUdict). `rlx-tiny-tts` / `rlx-melotts` materialize sidecars then run the ONNX graphs.
+Outer RLXPFLAT bundle: nested native subgraph packs under `graphs/` (hot weight tensors + graph IR sidecars) plus English frontend assets. No `.onnx` on Hub. `rlx-tiny-tts` materializes the outer pack, then lowers each nested pack to HIR for the utterance length.
 
 Official RLX package format (`RLXPFLAT`, container v2).
 
@@ -72,14 +72,14 @@ directly (or materialize sidecars to a temp dir for asset-only packs).
 
 ### Sidecars (file assets)
 
-Entire runtime tree is stored as cold zstd sidecars.
+Outer TOC is file sidecars only. Neural weights live **inside** each `graphs/<name>.rlxp` as hot mmap tensors (not safetensors).
 
 Paths below are logical ids inside the pack (`__flat__/sidecar/<id>`).
 Cold sidecars are zstd-compressed; sizes show raw → stored.
 
 | Sidecar | Raw | Stored | Role |
 |---|---:|---:|---|
-| `config.json` | 325 B | 220 B | Model / sample-rate config |
+| `config.json` | 325 B | 220 B | Model / sample-rate config (44.1 kHz) |
 | `frontend/bert/tokenizer.json` | 694.7 KiB | 214.5 KiB | BERT WordPiece tokenizer |
 | `frontend/bert/tokenizer_config.json` | 351 B | 218 B | Tokenizer config |
 | `frontend/cmudict_rep.txt` | 3.9 MiB | 1.0 MiB | CMUdict repair / variants |
@@ -88,36 +88,48 @@ Cold sidecars are zstd-compressed; sizes show raw → stored.
 | `frontend/homographs.en` | 18.3 KiB | 5.0 KiB | Homograph list |
 | `frontend/perceptron_tagger.json` | 5.4 MiB | 1.5 MiB | POS tagger |
 | `frontend/symbols.json` | 2.1 KiB | 726 B | Symbol table |
-| `onnx/decoder.onnx` | 13.8 MiB | 12.5 MiB | VITS2 decoder |
-| `onnx/duration_predictor.onnx` | 1.0 MiB | 598.4 KiB | Duration predictor |
-| `onnx/flow.onnx` | 14.2 MiB | 7.9 MiB | Flow / posterior |
-| `onnx/text_encoder.onnx` | 3.0 MiB | 1.7 MiB | Text encoder |
+| `graphs/decoder.rlxp` | 13.7 MiB | 12.5 MiB | Nested pack: hot tensors + graph.json |
+| `graphs/duration_predictor.rlxp` | 1.0 MiB | 598.8 KiB | Nested pack: hot tensors + graph.json |
+| `graphs/flow.rlxp` | 13.7 MiB | 8.0 MiB | Nested pack: hot tensors + graph.json |
+| `graphs/text_encoder.rlxp` | 2.9 MiB | 1.7 MiB | Nested pack: hot tensors + graph.json |
+
+### Architecture
+
+**Pipeline:** text → English frontend (G2P) → `text_encoder` → `duration_predictor` → monotonic alignment + latent sample (Rust) → `flow` → `decoder` → 44.1 kHz mono WAV.
+
+| Module | Role | Typical dtype | Notes |
+|---|---|---|---|
+| `text_encoder` | phone/tone/lang → latent | f32 | length = phoneme count `T` |
+| `duration_predictor` | per-phone duration | f32 | |
+| `flow` | prior / flow | f32 | |
+| `decoder` | HiFi-GAN-style vocoder | f32 | ConvTranspose upsample ×512 |
+| frontend | G2P + BERT tokenizer + CMUdict | — | file sidecars |
 
 ### Logical tree
 
 ```text
 tiny-tts.rlxp
-└── sidecars/
-    ├── config.json
-    ├── onnx/
-    │   ├── text_encoder.onnx
-    │   ├── duration_predictor.onnx
-    │   ├── flow.onnx
-    │   └── decoder.onnx
-    └── frontend/
-        ├── bert/tokenizer*.json
-        ├── g2p_*.{safetensors,txt}
-        └── symbols / tagger / homographs
+├── graphs/
+│   ├── text_encoder.rlxp      # hot f32/i64 + graph.json
+│   ├── duration_predictor.rlxp
+│   ├── flow.rlxp
+│   └── decoder.rlxp
+├── config.json
+└── frontend/                 # G2P / tokenizer (not neural)
 ```
 
 ### How it is packed
 
-`just export-tiny-tts-rlxp` → `rlx-assets` `write_dir_rlxp` over `weights/tts/tiny-tts-rlx` (ONNX + frontend tree).
+`just export-tiny-tts-rlxp` / `pack_rlxp` example: ONNX pack-time source → nested `graphs/*.rlxp` via `rlx-assets` native-pack, then outer bundle. Hub artifact has zero `.onnx`.
 
 ## Note
 
 MeloTTS (`rlx-melotts`) loads this same bundle — locally `weights/tts/melotts` is a symlink.
 
+## Note
+
+Hub ships **no ONNX**. Nested `graphs/*.rlxp` hold hot f32/i64 tensors + `graph.json`; the crate lowers to HIR per utterance length.
+
 ## Run with RLX
 
 Clone [rlx-models](https://github.com/MIT-RLX/rlx-models), place this repo under `weights/tts/tiny-tts-rlx` (or pass the path explicitly), then:

tiny-tts.rlxp

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:90627cd76af165a9707d368edab801aa91d654dc7c644267702ce37550f837db
-size 30811106
+oid sha256:5a8c8ea4b10be5d5fb960e5c8e8ebcbc61c8c32e3fbf9f094ec3d2b22d066e73
+size 30835846