Parametrising fast inference so that finetuned models can be used (me…

…tavoiceio#113)

README.md

@@ -6,7 +6,7 @@
 <a target="_blank" style="display: inline-block; vertical-align: middle" href="https://colab.research.google.com/github/metavoiceio/metavoice-src/blob/main/colab_demo.ipynb">
   <img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open In Colab"/>
 </a>
-[![](https://dcbadge.vercel.app/api/server/Cpy6U3na8Z?style=flat&compact=True)](https://discord.gg/tbTbkGEgJM) 
+[![](https://dcbadge.vercel.app/api/server/Cpy6U3na8Z?style=flat&compact=True)](https://discord.gg/tbTbkGEgJM)
 [![Twitter](https://img.shields.io/twitter/url/https/twitter.com/OnusFM.svg?style=social&label=@metavoiceio)](https://twitter.com/metavoiceio)
 
 
@@ -69,7 +69,7 @@ poetry install && poetry run pip install torch==2.2.1 torchaudio==2.2.1
 ## Usage
 1. Download it and use it anywhere (including locally) with our [reference implementation](/fam/llm/fast_inference.py)
 ```bash
-# You can use `--quantisation_mode int4` or `--quantisation_mode int8` for experimental faster inference.  This will degrade the quality of the audio. 
+# You can use `--quantisation_mode int4` or `--quantisation_mode int8` for experimental faster inference.  This will degrade the quality of the audio.
 # Note: int8 is slower than bf16/fp16 for undebugged reasons. If you want fast, try int4 which is roughly 2x faster than bf16/fp16.
 poetry run python -i fam/llm/fast_inference.py
 
@@ -82,7 +82,7 @@ tts.synthesise(text="This is a demo of text to speech by MetaVoice-1B, an open-s
 
 2. Deploy it on any cloud (AWS/GCP/Azure), using our [inference server](serving.py) or [web UI](app.py)
 ```bash
-# You can use `--quantisation_mode int4` or `--quantisation_mode int8` for experimental faster inference. This will degrade the quality of the audio. 
+# You can use `--quantisation_mode int4` or `--quantisation_mode int8` for experimental faster inference. This will degrade the quality of the audio.
 # Note: int8 is slower than bf16/fp16 for undebugged reasons. If you want fast, try int4 which is roughly 2x faster than bf16/fp16.
 poetry run python serving.py
 poetry run python app.py
@@ -108,6 +108,11 @@ Try it out using our sample datasets via:
 poetry run finetune --train ./datasets/sample_dataset.csv --val ./datasets/sample_val_dataset.csv
 ```
 
+Once you've trained your model, you can use it for inference via:
+```bash
+poetry run python -i fam/llm/fast_inference.py --first_stage_path ./my-finetuned_model.pt
+```
+
 ### Configuration
 
 In order to set hyperparameters such as learning rate, what to freeze, etc, you

fam/llm/fast_inference.py

@@ -41,6 +41,7 @@ def __init__(
         seed: int = 1337,
         output_dir: str = "outputs",
         quantisation_mode: Optional[Literal["int4", "int8"]] = None,
+        first_stage_path: Optional[str] = None,
     ):
         """
         Initialise the TTS model.
@@ -54,6 +55,7 @@ def __init__(
                 - None for no quantisation (bf16 or fp16 based on device),
                 - int4 for int4 weight-only quantisation,
                 - int8 for int8 weight-only quantisation.
+            first_stage_path: path to first-stage LLM checkpoint. If provided, this will override the one grabbed from Hugging Face via `model_name`.
         """
 
         # NOTE: this needs to come first so that we don't change global state when we want to use
@@ -64,6 +66,9 @@ def __init__(
         self.first_stage_adapter = FlattenedInterleavedEncodec2Codebook(end_of_audio_token=self.END_OF_AUDIO_TOKEN)
         self.output_dir = output_dir
         os.makedirs(self.output_dir, exist_ok=True)
+        if first_stage_path:
+            print(f"Overriding first stage checkpoint via provided model: {first_stage_path}")
+        first_stage_ckpt = first_stage_path or f"{self._model_dir}/first_stage.pt"
 
         second_stage_ckpt_path = f"{self._model_dir}/second_stage.pt"
         config_second_stage = InferenceConfig(
@@ -85,7 +90,7 @@ def __init__(
         self.precision = {"float16": torch.float16, "bfloat16": torch.bfloat16}[self._dtype]
         self.model, self.tokenizer, self.smodel, self.model_size = build_model(
             precision=self.precision,
-            checkpoint_path=Path(f"{self._model_dir}/first_stage.pt"),
+            checkpoint_path=Path(first_stage_ckpt),
             spk_emb_ckpt_path=Path(f"{self._model_dir}/speaker_encoder.pt"),
             device=self._device,
             compile=True,