From 9161ad080ce791e63286f599332d1d2b33153ab3 Mon Sep 17 00:00:00 2001
From: Zhaoheng Ni <zni@meta.com>
Date: Wed, 26 Jun 2024 08:37:23 -0700
Subject: [PATCH] Fix model downloading in bento

Summary: The model checkpoint path can not be created for Squim models. Use the latest download_asset method to fix it.

Differential Revision: D59061348
---
 src/torchaudio/pipelines/_squim_pipeline.py | 40 ++++++---------------
 1 file changed, 10 insertions(+), 30 deletions(-)

diff --git a/src/torchaudio/pipelines/_squim_pipeline.py b/src/torchaudio/pipelines/_squim_pipeline.py
index f1d11bff538..0c70db4aef7 100644
--- a/src/torchaudio/pipelines/_squim_pipeline.py
+++ b/src/torchaudio/pipelines/_squim_pipeline.py
@@ -1,6 +1,7 @@
 from dataclasses import dataclass
 
-from torchaudio._internal import load_state_dict_from_url
+import torch
+import torchaudio
 
 from torchaudio.models import squim_objective_base, squim_subjective_base, SquimObjective, SquimSubjective
 
@@ -42,26 +43,16 @@ class SquimObjectiveBundle:
     _path: str
     _sample_rate: float
 
-    def _get_state_dict(self, dl_kwargs):
-        url = f"https://download.pytorch.org/torchaudio/models/{self._path}"
-        dl_kwargs = {} if dl_kwargs is None else dl_kwargs
-        state_dict = load_state_dict_from_url(url, **dl_kwargs)
-        return state_dict
-
-    def get_model(self, *, dl_kwargs=None) -> SquimObjective:
+    def get_model(self) -> SquimObjective:
         """Construct the SquimObjective model, and load the pretrained weight.
 
-        The weight file is downloaded from the internet and cached with
-        :func:`torch.hub.load_state_dict_from_url`
-
-        Args:
-            dl_kwargs (dictionary of keyword arguments): Passed to :func:`torch.hub.load_state_dict_from_url`.
-
         Returns:
             Variation of :py:class:`~torchaudio.models.SquimObjective`.
         """
         model = squim_objective_base()
-        model.load_state_dict(self._get_state_dict(dl_kwargs))
+        path = torchaudio.utils.download_asset(f"models/{self._path}")
+        state_dict = torch.load(path, weights_only=True)
+        model.load_state_dict(state_dict)
         model.eval()
         return model
 
@@ -128,26 +119,15 @@ class SquimSubjectiveBundle:
     _path: str
     _sample_rate: float
 
-    def _get_state_dict(self, dl_kwargs):
-        url = f"https://download.pytorch.org/torchaudio/models/{self._path}"
-        dl_kwargs = {} if dl_kwargs is None else dl_kwargs
-        state_dict = load_state_dict_from_url(url, **dl_kwargs)
-        return state_dict
-
-    def get_model(self, *, dl_kwargs=None) -> SquimSubjective:
+    def get_model(self) -> SquimSubjective:
         """Construct the SquimSubjective model, and load the pretrained weight.
-
-        The weight file is downloaded from the internet and cached with
-        :func:`torch.hub.load_state_dict_from_url`
-
-        Args:
-            dl_kwargs (dictionary of keyword arguments): Passed to :func:`torch.hub.load_state_dict_from_url`.
-
         Returns:
             Variation of :py:class:`~torchaudio.models.SquimObjective`.
         """
         model = squim_subjective_base()
-        model.load_state_dict(self._get_state_dict(dl_kwargs))
+        path = torchaudio.utils.download_asset(f"models/{self._path}")
+        state_dict = torch.load(path, weights_only=True)
+        model.load_state_dict(state_dict)
         model.eval()
         return model