Add requirements

app.py

@@ -5,7 +5,6 @@ import mdtex2html
 import tempfile
 from PIL import Image
 import scipy
-import argparse
 
 from llama.m2ugen import M2UGen
 import llama
@@ -18,44 +17,9 @@ import av
 import subprocess
 import librosa
 
-parser = argparse.ArgumentParser()
-parser.add_argument(
-    "--model", default="./ckpts/checkpoint.pth", type=str,
-    help="Name of or path to M2UGen pretrained checkpoint",
-)
-parser.add_argument(
-    "--llama_type", default="7B", type=str,
-    help="Type of llama original weight",
-)
-parser.add_argument(
-    "--llama_dir", default="/path/to/llama", type=str,
-    help="Path to LLaMA pretrained checkpoint",
-)
-parser.add_argument(
-    "--mert_path", default="m-a-p/MERT-v1-330M", type=str,
-    help="Path to MERT pretrained checkpoint",
-)
-parser.add_argument(
-    "--vit_path", default="m-a-p/MERT-v1-330M", type=str,
-    help="Path to ViT pretrained checkpoint",
-)
-parser.add_argument(
-    "--vivit_path", default="m-a-p/MERT-v1-330M", type=str,
-    help="Path to ViViT pretrained checkpoint",
-)
-parser.add_argument(
-    "--knn_dir", default="./ckpts", type=str,
-    help="Path to directory with KNN Index",
-)
-parser.add_argument(
-    '--music_decoder', default="musicgen", type=str,
-    help='Decoder to use musicgen/audioldm2')
-
-parser.add_argument(
-    '--music_decoder_path', default="facebook/musicgen-medium", type=str,
-    help='Path to decoder to use musicgen/audioldm2')
-
-args = parser.parse_args()
+args = {"model": "./ckpts/M2UGen/checkpoint.pth", "llama_type": "7B", "llama_dir": "./ckpts/LLaMA-2",
+        "mert_path": "m-a-p/MERT-v1-330M", "vit_path": "google/vit-base-patch16-224", "vivit_path": "google/vivit-b-16x2-kinetics400",
+        "music_decoder": "musicgen", "music_decoder_path": "facebook/musicgen-medium"}
 
 generated_audio_files = []
 
@@ -78,10 +42,6 @@ load_result = model.load_state_dict(new_ckpt, strict=False)
 assert len(load_result.unexpected_keys) == 0, f"Unexpected keys: {load_result.unexpected_keys}"
 model.eval()
 model.to("cuda")
-#model.generation_model.to("cuda")
-#model.mert_model.to("cuda")
-#model.vit_model.to("cuda")
-#model.vivit_model.to("cuda")
 
 transform = transforms.Compose(
     [transforms.ToTensor(), transforms.Lambda(lambda x: x.repeat(3, 1, 1) if x.size(0) == 1 else x)])