update

app.py

@@ -79,6 +79,7 @@ if not os.path.exists(ckpt_dir):
 
 snacmodel = SNAC.from_pretrained("hubertsiuzdak/snac_24khz").eval().to(device)
 whispermodel = whisper.load_model("small").to(device)
+whispermodel.eval()
 text_tokenizer = Tokenizer(ckpt_dir)
 # fabric = L.Fabric(devices=1, strategy="auto")
 config = Config.from_file(ckpt_dir + "/model_config.yaml")
@@ -94,10 +95,10 @@ model.eval()
 
 @spaces.GPU
 def get_input_ids_whisper_ATBatch(mel, leng, whispermodel, device):
-    with torch.no_grad():
-        mel = mel.unsqueeze(0).to(device)
-        # audio_feature = whisper.decode(whispermodel,mel, options).audio_features
-        audio_feature = whispermodel.embed_audio(mel)[0][:leng]
+    # with torch.no_grad():
+    mel = mel.unsqueeze(0).to(device)
+    # audio_feature = whisper.decode(whispermodel,mel, options).audio_features
+    audio_feature = whispermodel.embed_audio(mel)[0][:leng]
     T = audio_feature.size(0)
     input_ids_AA = []
     for i in range(7):