1.try using git clone for structure and use wget for critical lfs files

download_models.py

@@ -11,15 +11,15 @@ from pathlib import Path
 
 
 def download_xcodec_models():
-    """Download xcodec_mini_infer models using git clone (supports Git LFS)"""
+    """Download xcodec_mini_infer using git clone (no LFS) + wget for large files"""
 
     # Base path for xcodec models
     xcodec_base = Path("YuEGP/inference/xcodec_mini_infer")
 
-    print("📥 Downloading xcodec_mini_infer models using git clone...")
+    print("📥 Downloading xcodec_mini_infer using git clone + wget strategy...")
 
     try:
-        # Remove existing directory if it exists (might be empty or incomplete)
+        # Remove existing directory if it exists
         if xcodec_base.exists():
             print("🗑️ Removing existing xcodec_mini_infer directory...")
             shutil.rmtree(xcodec_base)
@@ -32,39 +32,98 @@ def download_xcodec_models():
         os.chdir(xcodec_base.parent)
 
         try:
-            # Clone the repository using git (Spaces has git and git-lfs pre-installed)
-            print("🔄 Cloning repository with Git LFS support...")
-            result = subprocess.run([
-                "git", "clone",
+            # Step 1: Clone repository structure without LFS files
+            print("🔄 Step 1: Cloning repository structure (no LFS)...")
+            subprocess.run([
+                "git", "clone", "--no-lfs",
                 "https://huggingface.co/m-a-p/xcodec_mini_infer",
                 "xcodec_mini_infer"
-            ], check=True, capture_output=True, text=True, timeout=600)
-
-            print("✅ Successfully cloned xcodec_mini_infer repository")
+            ], check=True, capture_output=True, text=True, timeout=300)
+
+            print("✅ Repository structure downloaded successfully")
+
+            # Step 2: Download critical LFS files using wget
+            print("🔄 Step 2: Downloading critical LFS files with wget...")
+
+            # Define critical LFS files and their download URLs
+            lfs_files = [
+                {
+                    "path": "decoders/decoder_131000.pth",
+                    "url": "https://huggingface.co/m-a-p/xcodec_mini_infer/resolve/main/decoders/decoder_131000.pth",
+                    "description": "Vocal decoder model (70MB)"
+                },
+                {
+                    "path": "decoders/decoder_151000.pth",
+                    "url": "https://huggingface.co/m-a-p/xcodec_mini_infer/resolve/main/decoders/decoder_151000.pth",
+                    "description": "Instrumental decoder model (70MB)"
+                },
+                {
+                    "path": "final_ckpt/ckpt_00360000.pth",
+                    "url": "https://huggingface.co/m-a-p/xcodec_mini_infer/resolve/main/final_ckpt/ckpt_00360000.pth",
+                    "description": "Main checkpoint (1.3GB)"
+                },
+                {
+                    "path": "semantic_ckpts/hf_1_325000/pytorch_model.bin",
+                    "url": "https://huggingface.co/m-a-p/xcodec_mini_infer/resolve/main/semantic_ckpts/hf_1_325000/pytorch_model.bin",
+                    "description": "Semantic model (361MB)"
+                }
+            ]
 
-            # Verify critical decoder files exist (these are Git LFS files)
+            # Change to the cloned directory
+            os.chdir("xcodec_mini_infer")
+
+            # Download each LFS file
+            success_count = 0
+            for file_info in lfs_files:
+                try:
+                    file_path = Path(file_info["path"])
+                    print(f"📥 Downloading {file_info['description']}...")
+
+                    # Ensure directory exists
+                    os.makedirs(file_path.parent, exist_ok=True)
+
+                    # Download with wget
+                    subprocess.run([
+                        "wget", "-O", str(file_path),
+                        file_info["url"]
+                    ], check=True, capture_output=True, text=True, timeout=1800)  # 30min timeout for large files
+
+                    # Verify download
+                    if file_path.exists() and file_path.stat().st_size > 1024:  # > 1KB
+                        print(f"✅ Successfully downloaded {file_info['path']} ({file_path.stat().st_size // (1024*1024)}MB)")
+                        success_count += 1
+                    else:
+                        print(f"⚠️ {file_info['path']} download appears incomplete")
+
+                except subprocess.CalledProcessError as e:
+                    print(f"❌ Failed to download {file_info['path']}: {e}")
+                    if e.stderr:
+                        print(f"Error details: {e.stderr[-500:]}...")  # Last 500 chars
+                except subprocess.TimeoutExpired:
+                    print(f"❌ Download timeout for {file_info['path']} (large file)")
+                except Exception as e:
+                    print(f"❌ Unexpected error downloading {file_info['path']}: {e}")
+
+            os.chdir("..")
+
+            # Verify critical decoder files (minimum requirement)
             decoder_files = [
                 xcodec_base / "decoders" / "decoder_131000.pth",
                 xcodec_base / "decoders" / "decoder_151000.pth"
             ]
 
-            missing_decoders = [f for f in decoder_files if not f.exists()]
-            if missing_decoders:
-                print(f"⚠️ Decoder files missing: {[f.name for f in missing_decoders]}")
-                print("🔄 Ensuring Git LFS files are downloaded...")
+            missing_decoders = []
+            for decoder_file in decoder_files:
+                if not decoder_file.exists() or decoder_file.stat().st_size < 1024:
+                    missing_decoders.append(decoder_file.name)
 
-                # Try to explicitly pull LFS files
-                os.chdir("xcodec_mini_infer")
-                subprocess.run(["git", "lfs", "pull"], check=True, capture_output=True, text=True, timeout=300)
-                os.chdir("..")
-
-                # Check again
-                missing_decoders = [f for f in decoder_files if not f.exists()]
-                if missing_decoders:
-                    print(f"❌ Critical decoder files still missing: {[f.name for f in missing_decoders]}")
-                    return False
+            if missing_decoders:
+                print(f"❌ Critical decoder files missing or incomplete: {missing_decoders}")
+                print("Vocoder functionality will not work without decoder files.")
+                return False
 
-            print("✅ All critical model files verified present")
+            print(f"✅ Successfully downloaded {success_count}/{len(lfs_files)} LFS files")
+            print("✅ All critical decoder files verified present and complete")
             return True
 
         finally:
@@ -79,11 +138,11 @@ def download_xcodec_models():
         return False
 
     except subprocess.TimeoutExpired:
-        print("❌ Git clone timed out (network issue or large repository)")
+        print("❌ Git clone timed out")
         return False
 
     except Exception as e:
-        print(f"❌ Unexpected error during download: {e}")
+        print(f"❌ Unexpected error: {e}")
         return False