added additional adjustments and tests

app.py

@@ -130,6 +130,10 @@ def run(
     system_prompt: str,
     max_new_tokens: int,
     max_images: int,
+    temperature: float,
+    top_p: float,
+    top_k: int,
+    repetition_penalty: float,
 ) -> Iterator[str]:
 
     logger.debug(
@@ -162,6 +166,11 @@ def run(
         inputs,
         streamer=streamer,
         max_new_tokens=max_new_tokens,
+        temperature=temperature,
+        top_p=top_p,
+        top_k=top_k,
+        repetition_penalty=repetition_penalty,
+        do_sample=True,
     )
     t = Thread(target=model.generate, kwargs=generate_kwargs)
     t.start()
@@ -186,6 +195,18 @@ demo = gr.ChatInterface(
             label="Max New Tokens", minimum=100, maximum=2000, step=10, value=700
         ),
         gr.Slider(label="Max Images", minimum=1, maximum=4, step=1, value=2),
+        gr.Slider(
+            label="Temperature", minimum=0.1, maximum=2.0, step=0.1, value=0.7
+        ),
+        gr.Slider(
+            label="Top P", minimum=0.1, maximum=1.0, step=0.05, value=0.9
+        ),
+        gr.Slider(
+            label="Top K", minimum=1, maximum=100, step=1, value=50
+        ),
+        gr.Slider(
+            label="Repetition Penalty", minimum=1.0, maximum=2.0, step=0.05, value=1.1
+        )
     ],
     stop_btn=False,
 )

tests/test_media.py

@@ -5,7 +5,7 @@ from PIL import Image
 from pathlib import Path
 import tempfile
 
-from src.app import get_frames, process_video, process_user_input, process_history
+from app import get_frames, process_video, process_user_input, process_history
 
 # Get the project root directory
 ROOT_DIR = Path(__file__).parent.parent
@@ -235,135 +235,65 @@ def test_process_user_input_max_images_effect():
     assert frames_many <= 5
     assert frames_few < frames_many
     
-def test_empty_history():
-    """Test processing empty history."""
-    history = []
-    result = process_history(history)
-    assert result == []
-
-def test_single_user_message_text():
-    """Test processing a single user text message."""
-    history = [
-        {"role": "user", "content": "Hello, AI!"}
-    ]
+def test_process_history_basic_functionality():
+    """Test basic conversation processing and content buffering."""
+    # Empty history
+    assert process_history([]) == []
     
-    result = process_history(history)
-    
-    assert len(result) == 1
-    assert result[0]["role"] == "user"
-    assert len(result[0]["content"]) == 1
-    assert result[0]["content"][0]["type"] == "text"
-    assert result[0]["content"][0]["text"] == "Hello, AI!"
-
-def test_single_user_message_image():
-    """Test processing a single user image message."""
-    history = [
-        {"role": "user", "content": ["path/to/image.jpg"]}
-    ]
-    
-    result = process_history(history)
-    
-    assert len(result) == 1
-    assert result[0]["role"] == "user"
-    assert len(result[0]["content"]) == 1
-    assert result[0]["content"][0]["type"] == "image"
-    assert result[0]["content"][0]["url"] == "path/to/image.jpg"
-
-def test_single_assistant_message():
-    """Test processing a single assistant message."""
-    history = [
-        {"role": "assistant", "content": "I'm an AI assistant."}
-    ]
-    
-    result = process_history(history)
-    
-    assert len(result) == 1
-    assert result[0]["role"] == "assistant"
-    assert len(result[0]["content"]) == 1
-    assert result[0]["content"][0]["type"] == "text"
-    assert result[0]["content"][0]["text"] == "I'm an AI assistant."
-
-def test_alternating_messages():
-    """Test processing alternating user and assistant messages."""
+    # Simple conversation
     history = [
         {"role": "user", "content": "Hello"},
-        {"role": "assistant", "content": "Hi there"},
-        {"role": "user", "content": "How are you?"},
-        {"role": "assistant", "content": "I'm doing well, thanks!"}
-    ]
-    
-    result = process_history(history)
-    
-    assert len(result) == 4
-    assert [item["role"] for item in result] == ["user", "assistant", "user", "assistant"]
-    assert result[0]["content"][0]["text"] == "Hello"
-    assert result[1]["content"][0]["text"] == "Hi there"
-    assert result[2]["content"][0]["text"] == "How are you?"
-    assert result[3]["content"][0]["text"] == "I'm doing well, thanks!"
-
-def test_consecutive_user_messages():
-    """Test processing consecutive user messages - they should be grouped."""
-    history = [
-        {"role": "user", "content": "First message"},
-        {"role": "user", "content": "Second message"},
-        {"role": "user", "content": "Third message"},
-        {"role": "assistant", "content": "I got your messages"}
+        {"role": "assistant", "content": "Hi there!"},
+        {"role": "user", "content": "How are you?"}
     ]
     
     result = process_history(history)
-    
-    # Should be combined into a single user message with multiple content items
-    assert len(result) == 2
-    assert result[0]["role"] == "user"
-    assert len(result[0]["content"]) == 3
-    assert [item["text"] for item in result[0]["content"]] == [
-        "First message", "Second message", "Third message"
-    ]
-
-def test_mixed_content_types():
-    """Test processing mixed content types (text and images)."""
-    history = [
-        {"role": "user", "content": "Look at this:"},
-        {"role": "user", "content": ["image.jpg"]},
-        {"role": "assistant", "content": "Nice image!"}
-    ]
-    
-    result = process_history(history)
-    
-    assert len(result) == 2
-    assert result[0]["role"] == "user"
-    assert len(result[0]["content"]) == 2
-    assert result[0]["content"][0]["type"] == "text"
-    assert result[0]["content"][1]["type"] == "image"
-
-def test_ending_with_user_messages():
-    """Test history that ends with user messages."""
-    history = [
-        {"role": "user", "content": "Hello"},
-        {"role": "assistant", "content": "Hi there"},
-        {"role": "user", "content": "Another question"},
-        {"role": "user", "content": ["image.png"]}
-    ]
-    
-    result = process_history(history)
-    
     assert len(result) == 3
-    assert result[2]["role"] == "user"
-    assert len(result[2]["content"]) == 2
-    assert result[2]["content"][0]["type"] == "text"
-    assert result[2]["content"][1]["type"] == "image"
+    assert result[0] == {"role": "user", "content": [{"type": "text", "text": "Hello"}]}
+    assert result[1] == {"role": "assistant", "content": [{"type": "text", "text": "Hi there!"}]}
+    assert result[2] == {"role": "user", "content": [{"type": "text", "text": "How are you?"}]}
 
-def test_empty_messages():
-    """Test handling of empty content messages."""
-    history = [
-        {"role": "user", "content": ""},
-        {"role": "assistant", "content": ""},
-        {"role": "user", "content": "Hello"}
-    ]
+
+def test_process_history_file_handling():
+    """Test processing of different file types and content buffering."""
+    # Create temp image file
+    with tempfile.NamedTemporaryFile(suffix=".jpg", delete=False) as img:
+        image_path = img.name
     
-    result = process_history(history)
+    video_path = os.path.join(ROOT_DIR, "assets", "test_video.mp4")
     
-    assert len(result) == 3
-    assert result[0]["content"][0]["text"] == ""
-    assert result[1]["content"][0]["text"] == ""
-    assert result[2]["content"][0]["text"] == "Hello"
+    try:
+        history = [
+            {"role": "user", "content": (image_path,)},
+            {"role": "user", "content": "What's this image?"},
+            {"role": "user", "content": (video_path,)},
+            {"role": "assistant", "content": "I see an image and video."},
+            {"role": "user", "content": "First"},
+            {"role": "user", "content": "Second"},
+            {"role": "user", "content": "Third"}  # Multiple user messages at end
+        ]
+        
+        result = process_history(history)
+        assert len(result) == 3
+        
+        # First user turn: image + text + video
+        assert result[0]["role"] == "user"
+        assert len(result[0]["content"]) == 3
+        assert result[0]["content"][0] == {"type": "image", "url": image_path}
+        assert result[0]["content"][1] == {"type": "text", "text": "What's this image?"}
+        assert result[0]["content"][2] == {"type": "text", "text": "[Video uploaded previously]"}
+        
+        # Assistant response
+        assert result[1]["role"] == "assistant"
+        assert result[1]["content"] == [{"type": "text", "text": "I see an image and video."}]
+        
+        # Final user turn: multiple buffered messages
+        assert result[2]["role"] == "user"
+        assert len(result[2]["content"]) == 3
+        assert result[2]["content"][0] == {"type": "text", "text": "First"}
+        assert result[2]["content"][1] == {"type": "text", "text": "Second"}
+        assert result[2]["content"][2] == {"type": "text", "text": "Third"}
+        
+    finally:
+        if os.path.exists(image_path):
+            os.unlink(image_path)