omnigenome 0.3.25a0__tar.gz → 0.3.26a0__tar.gz

{omnigenome-0.3.25a0 → omnigenome-0.3.26a0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: omnigenome
-Version: 0.3.25a0
+Version: 0.3.26a0
 Summary: OmniGenome: A comprehensive toolkit for genome analysis.
 Home-page: https://github.com/yangheng95/OmniGenBench
 Author: Yang, Heng
@@ -182,7 +182,7 @@ ogb autobench \
     --trainer accelerate
 
 # Legacy command (still supported for backward compatibility)
-# autobench --model_name_or_path "yangheng/OmniGenome-186M" --benchmark "RGB"
+# autobench --config_or_model "yangheng/OmniGenome-186M" --benchmark "RGB"
 ```
 **Output**: Results include mean ± standard deviation for each metric (e.g., MCC: 0.742 ± 0.015, F1: 0.863 ± 0.009)
 
@@ -202,7 +202,7 @@ seeds = [0, 1, 2, 3, 4]  # Multi-seed for statistical rigor
 # Run automated evaluation
 bench = AutoBench(
     benchmark=benchmark,
-    model_name_or_path=gfm,
+    config_or_model=gfm,
     overwrite=False  # Skip completed tasks
 )
 bench.run(autocast=False, batch_size=bench_size, seeds=seeds)

{omnigenome-0.3.25a0 → omnigenome-0.3.26a0}/omnigenome.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: omnigenome
-Version: 0.3.25a0
+Version: 0.3.26a0
 Summary: OmniGenome: A comprehensive toolkit for genome analysis.
 Home-page: https://github.com/yangheng95/OmniGenBench
 Author: Yang, Heng
@@ -182,7 +182,7 @@ ogb autobench \
     --trainer accelerate
 
 # Legacy command (still supported for backward compatibility)
-# autobench --model_name_or_path "yangheng/OmniGenome-186M" --benchmark "RGB"
+# autobench --config_or_model "yangheng/OmniGenome-186M" --benchmark "RGB"
 ```
 **Output**: Results include mean ± standard deviation for each metric (e.g., MCC: 0.742 ± 0.015, F1: 0.863 ± 0.009)
 
@@ -202,7 +202,7 @@ seeds = [0, 1, 2, 3, 4]  # Multi-seed for statistical rigor
 # Run automated evaluation
 bench = AutoBench(
     benchmark=benchmark,
-    model_name_or_path=gfm,
+    config_or_model=gfm,
     overwrite=False  # Skip completed tasks
 )
 bench.run(autocast=False, batch_size=bench_size, seeds=seeds)

{omnigenome-0.3.25a0 → omnigenome-0.3.26a0}/omnigenome.egg-info/SOURCES.txt

@@ -11,6 +11,7 @@ omnigenome.egg-info/top_level.txt
 tests/test_attention_extraction.py
 tests/test_autobench_autotrain.py
 tests/test_autoinfer_cli.py
+tests/test_benchmark_download.py
 tests/test_cli_commands.py
 tests/test_genomic_embeddings.py
 tests/test_hf_download.py

{omnigenome-0.3.25a0 → omnigenome-0.3.26a0}/tests/test_attention_extraction.py

@@ -46,7 +46,7 @@ class TestAttentionExtractionEmbeddingModel:
     @pytest.fixture(scope="class")
     def embedding_model(self, model_name):
         """Load embedding model for attention extraction"""
-        # OmniModelForEmbedding takes model_name_or_path as first positional argument
+        # OmniModelForEmbedding takes config_or_model as first positional argument
         model = OmniModelForEmbedding(model_name, trust_remote_code=True)
         return model
 

{omnigenome-0.3.25a0 → omnigenome-0.3.26a0}/tests/test_autobench_autotrain.py

@@ -60,7 +60,7 @@ class TestAutoBenchAPI:
         # Basic initialization
         bench = AutoBench(
             benchmark="RGB",
-            model_name_or_path="yangheng/OmniGenome-186M",
+            config_or_model="yangheng/OmniGenome-186M",
             overwrite=False
         )
         
@@ -72,7 +72,7 @@ class TestAutoBenchAPI:
         for benchmark in benchmark_names:
             bench = AutoBench(
                 benchmark=benchmark,
-                model_name_or_path="yangheng/OmniGenome-186M",
+                config_or_model="yangheng/OmniGenome-186M",
             )
             assert bench is not None
     
@@ -80,7 +80,7 @@ class TestAutoBenchAPI:
         """Test AutoBench accepts various configuration options"""
         bench = AutoBench(
             benchmark="RGB",
-            model_name_or_path="yangheng/OmniGenome-186M",
+            config_or_model="yangheng/OmniGenome-186M",
             tokenizer_name_or_path="yangheng/OmniGenome-186M",
             trainer="accelerate",
             overwrite=True,
@@ -98,7 +98,7 @@ class TestAutoBenchAPI:
         # Use smallest model and single seed for speed
         bench = AutoBench(
             benchmark="RGB",
-            model_name_or_path="yangheng/OmniGenome-52M",
+            config_or_model="yangheng/OmniGenome-52M",
             overwrite=True,
         )
         
@@ -121,7 +121,7 @@ class TestAutoBenchAPI:
         """Test AutoBench supports multi-seed evaluation"""
         bench = AutoBench(
             benchmark="RGB",
-            model_name_or_path="yangheng/OmniGenome-186M",
+            config_or_model="yangheng/OmniGenome-186M",
         )
         
         # Configuration with multiple seeds (as in examples)
@@ -147,7 +147,7 @@ class TestAutoTrainAPI:
         """Test AutoTrain can be initialized"""
         trainer = AutoTrain(
             dataset_name_or_path="translation_efficiency_prediction",
-            model_name_or_path="yangheng/PlantRNA-FM",
+            config_or_model="yangheng/PlantRNA-FM",
         )
         
         assert trainer is not None
@@ -159,7 +159,7 @@ class TestAutoTrainAPI:
         
         trainer = AutoTrain(
             dataset_name_or_path="translation_efficiency_prediction",
-            model_name_or_path="yangheng/PlantRNA-FM",
+            config_or_model="yangheng/PlantRNA-FM",
             output_dir=str(output_dir),
             num_labels=2,
             max_length=512,
@@ -188,7 +188,7 @@ class TestAutoTrainAPI:
         for config in task_configs:
             trainer = AutoTrain(
                 dataset_name_or_path=config["dataset"],
-                model_name_or_path="yangheng/OmniGenome-52M",
+                config_or_model="yangheng/OmniGenome-52M",
                 num_labels=config["num_labels"],
             )
             assert trainer is not None
@@ -204,7 +204,7 @@ class TestAutoTrainAPI:
         
         trainer = AutoTrain(
             dataset_name_or_path="translation_efficiency_prediction",
-            model_name_or_path="yangheng/PlantRNA-FM",
+            config_or_model="yangheng/PlantRNA-FM",
             output_dir=str(output_dir),
             epochs=1,
             batch_size=4,
@@ -387,7 +387,7 @@ class TestAutoWorkflowIntegration:
         # Step 1: Train a custom model
         trainer = AutoTrain(
             dataset_name_or_path="translation_efficiency_prediction",
-            model_name_or_path="yangheng/PlantRNA-FM",
+            config_or_model="yangheng/PlantRNA-FM",
             output_dir=str(output_dir),
             epochs=1,
             batch_size=4,
@@ -396,7 +396,7 @@ class TestAutoWorkflowIntegration:
         # Step 2: Benchmark the trained model
         bench = AutoBench(
             benchmark="RGB",
-            model_name_or_path=str(output_dir),
+            config_or_model=str(output_dir),
             overwrite=True,
         )
         
@@ -420,7 +420,7 @@ class TestAutoWorkflowIntegration:
         for model in models:
             bench = AutoBench(
                 benchmark=benchmark,
-                model_name_or_path=model,
+                config_or_model=model,
             )
             benches.append(bench)
         
@@ -448,7 +448,7 @@ class TestRealWorldBenchmarks:
         
         bench = AutoBench(
             benchmark="RGB",
-            model_name_or_path="yangheng/OmniGenome-52M",
+            config_or_model="yangheng/OmniGenome-52M",
             overwrite=True,
         )
         

omnigenome-0.3.26a0/tests/test_benchmark_download.py

@@ -0,0 +1,331 @@
+# -*- coding: utf-8 -*-
+# file: test_benchmark_download.py
+# time: 18:30 01/11/2025
+# author: YANG, HENG <hy345@exeter.ac.uk> (杨恒)
+# github: https://github.com/yangheng95
+# Copyright (C) 2019-2025. All Rights Reserved.
+
+"""
+Tests for robust benchmark and dataset downloading functionality.
+
+This module tests the enhanced download infrastructure that eliminates Git-LFS
+dependencies and provides automatic integrity verification.
+"""
+
+import os
+import shutil
+import pytest
+from pathlib import Path
+
+from omnigenbench.src.utility.hub_utils import download_benchmark
+from omnigenbench.src.utility.model_hub.hf_download import (
+    download_from_hf_hub,
+    verify_download_integrity,
+    list_hf_repo_files,
+    get_model_info,
+)
+
+
+class TestBenchmarkDownload:
+    """Test suite for benchmark downloading with HuggingFace Hub API."""
+
+    @pytest.fixture
+    def test_cache_dir(self, tmp_path):
+        """Create temporary cache directory for tests."""
+        cache_dir = tmp_path / "test_benchmarks"
+        cache_dir.mkdir(exist_ok=True)
+        yield str(cache_dir)
+        # Cleanup after test
+        if cache_dir.exists():
+            shutil.rmtree(cache_dir)
+
+    def test_download_benchmark_by_name(self, test_cache_dir):
+        """
+        Test downloading benchmark by short name with HF Hub API.
+        
+        Expected: Benchmark should be downloaded to cache directory.
+        """
+        try:
+            benchmark_path = download_benchmark(
+                "RGB",
+                use_hf_api=True,
+                cache_dir=test_cache_dir,
+                force_download=False,
+            )
+            
+            # Verify path exists
+            assert os.path.exists(benchmark_path), f"Benchmark path does not exist: {benchmark_path}"
+            
+            # Verify it's a directory
+            assert os.path.isdir(benchmark_path), f"Benchmark path is not a directory: {benchmark_path}"
+            
+            print(f"[SUCCESS] Benchmark downloaded to: {benchmark_path}")
+            
+        except Exception as e:
+            # If HF Hub doesn't have this benchmark, test passes with warning
+            pytest.skip(f"Benchmark not available on HF Hub: {e}")
+
+    def test_download_benchmark_from_hf_repo(self, test_cache_dir):
+        """
+        Test downloading benchmark from HuggingFace dataset repository.
+        
+        Expected: Benchmark should be downloaded using HF dataset identifier.
+        """
+        try:
+            benchmark_path = download_benchmark(
+                "yangheng/OmniGenBench_RGB",
+                use_hf_api=True,
+                cache_dir=test_cache_dir,
+            )
+            
+            assert os.path.exists(benchmark_path)
+            assert os.path.isdir(benchmark_path)
+            
+            print(f"[SUCCESS] Benchmark from HF repo: {benchmark_path}")
+            
+        except Exception as e:
+            pytest.skip(f"HF dataset not available: {e}")
+
+    def test_download_benchmark_force_redownload(self, test_cache_dir):
+        """
+        Test force re-downloading benchmark to update cache.
+        
+        Expected: Benchmark should be re-downloaded even if cached.
+        """
+        try:
+            # First download
+            benchmark_path1 = download_benchmark(
+                "RGB",
+                use_hf_api=True,
+                cache_dir=test_cache_dir,
+                force_download=False,
+            )
+            
+            # Force re-download
+            benchmark_path2 = download_benchmark(
+                "RGB",
+                use_hf_api=True,
+                cache_dir=test_cache_dir,
+                force_download=True,
+            )
+            
+            # Both should point to same location
+            assert benchmark_path1 == benchmark_path2
+            assert os.path.exists(benchmark_path2)
+            
+            print(f"[SUCCESS] Force re-download completed: {benchmark_path2}")
+            
+        except Exception as e:
+            pytest.skip(f"Force re-download test skipped: {e}")
+
+    def test_download_benchmark_fallback(self, test_cache_dir):
+        """
+        Test fallback to legacy HTTP download when HF Hub is unavailable.
+        
+        Expected: Should fall back to OmniGenome Space repository.
+        """
+        try:
+            # Try with use_hf_api=False to test fallback
+            benchmark_path = download_benchmark(
+                "RGB",
+                use_hf_api=False,  # Force legacy method
+                cache_dir=test_cache_dir,
+            )
+            
+            assert os.path.exists(benchmark_path)
+            assert os.path.isdir(benchmark_path)
+            
+            print(f"[SUCCESS] Fallback download completed: {benchmark_path}")
+            
+        except Exception as e:
+            pytest.skip(f"Fallback download test: {e}")
+
+
+class TestDatasetDownload:
+    """Test suite for dataset downloading via HuggingFace Hub API."""
+
+    @pytest.fixture
+    def test_cache_dir(self, tmp_path):
+        """Create temporary cache directory for tests."""
+        cache_dir = tmp_path / "test_datasets"
+        cache_dir.mkdir(exist_ok=True)
+        yield str(cache_dir)
+        if cache_dir.exists():
+            shutil.rmtree(cache_dir)
+
+    def test_download_dataset_from_hf_hub(self, test_cache_dir):
+        """
+        Test downloading dataset from HuggingFace Hub.
+        
+        Expected: Dataset should be downloaded successfully.
+        """
+        try:
+            dataset_path = download_from_hf_hub(
+                repo_id="yangheng/test_genomic_dataset",
+                repo_type="dataset",
+                cache_dir=test_cache_dir,
+                force_download=False,
+            )
+            
+            assert os.path.exists(dataset_path)
+            
+            # Verify integrity
+            is_valid = verify_download_integrity(dataset_path)
+            print(f"[INFO] Dataset integrity: {'Valid' if is_valid else 'Corrupted'}")
+            
+            print(f"[SUCCESS] Dataset downloaded to: {dataset_path}")
+            
+        except Exception as e:
+            pytest.skip(f"Dataset download test: {e}")
+
+    def test_selective_dataset_download(self, test_cache_dir):
+        """
+        Test selective file download for bandwidth optimization.
+        
+        Expected: Only specified file patterns should be downloaded.
+        """
+        try:
+            dataset_path = download_from_hf_hub(
+                repo_id="yangheng/test_genomic_dataset",
+                repo_type="dataset",
+                cache_dir=test_cache_dir,
+                allow_patterns=["*.json", "*.txt"],
+                ignore_patterns=["*.zip", "*.tar.gz"],
+            )
+            
+            assert os.path.exists(dataset_path)
+            
+            # Check that only allowed patterns exist
+            files = list(Path(dataset_path).rglob("*"))
+            json_txt_files = [f for f in files if f.suffix in [".json", ".txt"]]
+            zip_files = [f for f in files if f.suffix in [".zip", ".tar.gz"]]
+            
+            assert len(zip_files) == 0, "Ignored patterns should not be downloaded"
+            
+            print(f"[SUCCESS] Selective download: {len(json_txt_files)} files")
+            
+        except Exception as e:
+            pytest.skip(f"Selective download test: {e}")
+
+
+class TestMetadataQuery:
+    """Test suite for querying benchmark/dataset metadata."""
+
+    def test_list_repo_files(self):
+        """
+        Test listing files in a HuggingFace repository.
+        
+        Expected: Should return list of file paths.
+        """
+        try:
+            files = list_hf_repo_files(
+                "yangheng/OmniGenome-186M",
+                repo_type="model"
+            )
+            
+            assert isinstance(files, list), "Should return list of files"
+            assert len(files) > 0, "Should have at least some files"
+            
+            # Check for common model files
+            file_names = [os.path.basename(f) for f in files]
+            assert "config.json" in file_names, "Should have config.json"
+            
+            print(f"[SUCCESS] Found {len(files)} files in repository")
+            
+        except Exception as e:
+            pytest.skip(f"File listing test: {e}")
+
+    def test_get_model_info(self):
+        """
+        Test retrieving model repository metadata.
+        
+        Expected: Should return dict with metadata.
+        """
+        try:
+            info = get_model_info("yangheng/OmniGenome-186M")
+            
+            assert isinstance(info, dict), "Should return dictionary"
+            assert "id" in info or "modelId" in info, "Should have ID field"
+            
+            # Check for siblings (files)
+            if "siblings" in info:
+                total_size = sum(f.get("size", 0) for f in info["siblings"])
+                print(f"[INFO] Total model size: {total_size / (1024**2):.1f} MB")
+            
+            print(f"[SUCCESS] Retrieved metadata for model")
+            
+        except Exception as e:
+            pytest.skip(f"Metadata query test: {e}")
+
+
+class TestIntegrationWithAutoBench:
+    """Test integration of robust downloading with AutoBench."""
+
+    def test_autobench_automatic_download(self):
+        """
+        Test that AutoBench automatically uses robust downloading.
+        
+        Expected: Benchmark should be downloaded automatically when AutoBench is initialized.
+        """
+        try:
+            from omnigenbench import AutoBench
+            
+            # Initialize AutoBench - should automatically download benchmark
+            bench = AutoBench(
+                benchmark="RGB",
+                config_or_model="yangheng/OmniGenome-186M",
+                trainer="native",
+            )
+            
+            # Check that benchmark was downloaded
+            assert os.path.exists(bench.benchmark), f"Benchmark not downloaded: {bench.benchmark}"
+            assert os.path.isdir(bench.benchmark), "Benchmark should be a directory"
+            
+            # Check for metadata file
+            metadata_path = os.path.join(bench.benchmark, "metadata.py")
+            assert os.path.exists(metadata_path), "Benchmark should have metadata.py"
+            
+            print(f"[SUCCESS] AutoBench initialized with benchmark at: {bench.benchmark}")
+            
+        except Exception as e:
+            pytest.skip(f"AutoBench integration test: {e}")
+
+
+# Utility tests
+def test_benchmark_download_without_hf_api():
+    """
+    Test benchmark download with HF Hub API disabled (legacy method).
+    
+    Expected: Should use legacy HTTP download from OmniGenome Space.
+    """
+    try:
+        benchmark_path = download_benchmark(
+            "RGB",
+            use_hf_api=False,  # Disable HF Hub API
+        )
+        
+        assert os.path.exists(benchmark_path)
+        print(f"[SUCCESS] Legacy download method works: {benchmark_path}")
+        
+    except Exception as e:
+        pytest.skip(f"Legacy download test: {e}")
+
+
+def test_benchmark_download_with_invalid_name():
+    """
+    Test error handling for invalid benchmark names.
+    
+    Expected: Should raise ValueError or skip gracefully.
+    """
+    with pytest.raises((ValueError, Exception)):
+        download_benchmark(
+            "NonExistentBenchmark123456",
+            use_hf_api=True,
+        )
+    
+    print("[SUCCESS] Invalid benchmark name handled correctly")
+
+
+if __name__ == "__main__":
+    # Run tests with pytest
+    pytest.main([__file__, "-v", "-s"])