Update app.py

app.py

@@ -1,23 +1,29 @@
 import gradio as gr
 import torch
 from transformers import pipeline, AutoTokenizer
+from nemo.collections.asr.models import EncDecMultiTaskModel
 
-# tokenizer = AutoTokenizer.from_pretrained("microsoft/Phi-3-mini-128k-instruct")
+# load model
+canary_model = EncDecMultiTaskModel.from_pretrained('nvidia/canary-1b')
 
-# pipe = pipeline("text-generation", model="microsoft/Phi-3-mini-128k-instruct", trust_remote_code=True, tokenizer=tokenizer)
-# gr.Interface.from_pipeline(pipe,
-#                            title="TextGen",
-#                            description="Using pipeline with Phi3",
-#                            ).launch(inbrowser=True)
+# update dcode params
+decode_cfg = canary_model.cfg.decoding
+decode_cfg.beam.beam_size = 1
+canary_model.change_decoding_strategy(decode_cfg)
 
 pipe = pipeline(
-    "text-generation", 
-    model="QuantFactory/Meta-Llama-3-8B-Instruct-GGUF", 
-    model_kwargs={"torch_dtype": torch.bfloat16}, 
-    device_map="auto"
+    "automatic-speech-recognition", 
+    model="nvidia/canary-1b"
 )
 
+# pipe = pipeline(
+#     "text-generation", 
+#     model="QuantFactory/Meta-Llama-3-8B-Instruct-GGUF", 
+#     model_kwargs={"torch_dtype": torch.bfloat16}, 
+#     device_map="auto"
+# )
+
 gr.Interface.from_pipeline(pipe,
-                           title="TextGen",
-                           description="Using pipeline with Llama3",
+                           title="ASR",
+                           description="Using pipeline with Canary-1B",
                            ).launch(inbrowser=True)