webscout 7.0__py3-none-any.whl → 7.2__py3-none-any.whl

webscout/LLM.py

@@ -1,442 +1,442 @@
-"""
->>> from webscout.LLM import LLM, VLM
->>> llm = LLM("meta-llama/Meta-Llama-3-70B-Instruct")
->>> response = llm.chat([{"role": "user", "content": "What's good?"}])
->>> print(response)
-'Hey! I'm doing great, thanks for asking! How can I help you today? 😊'
-
->>> # For vision tasks
->>> vlm = VLM("cogvlm-grounding-generalist")
->>> response = vlm.chat([{"role": "user", "content": [{"type": "image", "image_url": "path/to/image.jpg"}, {"type": "text", "text": "What's in this image?"}]}])
-"""
-
-import requests
-import base64
-import json
-from typing import List, Dict, Union, Generator, Optional, Any
-
-class LLMError(Exception):
-    """Custom exception for LLM API errors 🚫
-
-    Examples:
-        >>> try:
-        ...     raise LLMError("API key not found!")
-        ... except LLMError as e:
-        ...     print(f"Error: {e}")
-        Error: API key not found!
-    """
-    pass
-
-class LLM:
-    """A class for chatting with DeepInfra's powerful language models! 🚀
-
-    This class lets you:
-    - Chat with state-of-the-art language models 💬
-    - Stream responses in real-time ⚡
-    - Control temperature and token limits 🎮
-    - Handle system messages and chat history 📝
-
-    Examples:
-        >>> from webscout.LLM import LLM
-        >>> llm = LLM("meta-llama/Meta-Llama-3-70B-Instruct")
-        >>> response = llm.chat([
-        ...     {"role": "user", "content": "Write a short poem!"}
-        ... ])
-        >>> print(response)
-        'Through starlit skies and morning dew,
-        Nature's beauty, forever new.
-        In every moment, magic gleams,
-        Life's poetry flows like gentle streams.'
-    """
-    
-    def __init__(self, model: str, system_message: str = "You are a Helpful AI."):
-        """
-        Initialize the LLM client.
-        
-        Args:
-            model: The model identifier (e.g., "meta-llama/Meta-Llama-3-70B-Instruct")
-            system_message: The system message to use for the conversation
-            
-        Examples:
-            >>> llm = LLM("meta-llama/Meta-Llama-3-70B-Instruct")
-            >>> print(llm.model)
-            'meta-llama/Meta-Llama-3-70B-Instruct'
-        """
-        self.model = model
-        self.api_url = "https://api.deepinfra.com/v1/openai/chat/completions"
-        self.conversation_history = [{"role": "system", "content": system_message}]
-        self.headers = {
-            'User-Agent': 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/119.0.0.0 Safari/537.36',
-            'Accept-Language': 'en,fr-FR;q=0.9,fr;q=0.8,es-ES;q=0.7,es;q=0.6,en-US;q=0.5,am;q=0.4,de;q=0.3',
-            'Cache-Control': 'no-cache',
-            'Connection': 'keep-alive',
-            'Content-Type': 'application/json',
-            'Origin': 'https://deepinfra.com',
-            'Pragma': 'no-cache',
-            'Referer': 'https://deepinfra.com/',
-            'Sec-Fetch-Dest': 'empty',
-            'Sec-Fetch-Mode': 'cors',
-            'Sec-Fetch-Site': 'same-site',
-            'X-Deepinfra-Source': 'web-embed',
-            'accept': 'text/event-stream',
-            'sec-ch-ua': '"Google Chrome";v="119", "Chromium";v="119", "Not?A_Brand";v="24"',
-            'sec-ch-ua-mobile': '?0',
-            'sec-ch-ua-platform': '"macOS"'
-        }
-
-    def _prepare_payload(
-        self, 
-        messages: List[Dict[str, str]], 
-        stream: bool = False,
-        temperature: float = 0.7,
-        max_tokens: int = 8028,
-        stop: Optional[List[str]] = None,
-    ) -> Dict[str, Any]:
-        """Prepare the chat payload with all the right settings! 🎯
-
-        Args:
-            messages: Your chat messages (role & content)
-            stream: Want real-time responses? Set True! ⚡
-            temperature: Creativity level (0-1) 🎨
-            max_tokens: Max words to generate 📝
-            stop: Words to stop at (optional) 🛑
-
-        Returns:
-            Dict with all the API settings ready to go! 🚀
-
-        Examples:
-            >>> llm = LLM("meta-llama/Meta-Llama-3-70B-Instruct")
-            >>> payload = llm._prepare_payload([
-            ...     {"role": "user", "content": "Hi!"}
-            ... ])
-            >>> print(payload['model'])
-            'meta-llama/Meta-Llama-3-70B-Instruct'
-        """
-        return {
-            'model': self.model,
-            'messages': messages,
-            'temperature': temperature,
-            'max_tokens': max_tokens,
-            'stop': stop or [],
-            'stream': stream
-        }
-
-    def chat(
-        self, 
-        messages: List[Dict[str, str]], 
-        stream: bool = False,
-        temperature: float = 0.7,
-        max_tokens: int = 8028,
-        stop: Optional[List[str]] = None,
-    ) -> Union[str, Generator[str, None, None]]:
-        """Start chatting with the AI! 💬
-
-        This method is your gateway to:
-        - Having awesome conversations 🗣️
-        - Getting creative responses 🎨
-        - Streaming real-time replies ⚡
-        - Controlling the output style 🎮
-
-        Args:
-            messages: Your chat messages (role & content)
-            stream: Want real-time responses? Set True!
-            temperature: Creativity level (0-1)
-            max_tokens: Max words to generate
-            stop: Words to stop at (optional)
-
-        Returns:
-            Either a complete response or streaming generator
-
-        Raises:
-            LLMError: If something goes wrong 🚫
-
-        Examples:
-            >>> llm = LLM("meta-llama/Meta-Llama-3-70B-Instruct")
-            >>> # Regular chat
-            >>> response = llm.chat([
-            ...     {"role": "user", "content": "Tell me a joke!"}
-            ... ])
-            >>> # Streaming chat
-            >>> for chunk in llm.chat([
-            ...     {"role": "user", "content": "Tell me a story!"}
-            ... ], stream=True):
-            ...     print(chunk, end='')
-        """
-        payload = self._prepare_payload(messages, stream, temperature, max_tokens, stop)
-        
-        try:
-            if stream:
-                return self._stream_response(payload)
-            else:
-                return self._send_request(payload)
-        except Exception as e:
-            raise LLMError(f"API request failed: {str(e)}")
-
-    def _stream_response(self, payload: Dict[str, Any]) -> Generator[str, None, None]:
-        """Stream the chat response in real-time! ⚡
-
-        Args:
-            payload: The prepared chat payload
-
-        Yields:
-            Streaming chunks of the response
-
-        Raises:
-            LLMError: If the stream request fails 🚫
-
-        Examples:
-            >>> llm = LLM("meta-llama/Meta-Llama-3-70B-Instruct")
-            >>> for chunk in llm._stream_response(llm._prepare_payload([
-            ...     {"role": "user", "content": "Tell me a story!"}
-            ... ])):
-            ...     print(chunk, end='')
-        """
-        try:
-            with requests.post(self.api_url, json=payload, headers=self.headers, stream=True) as response:
-                response.raise_for_status()
-                for line in response.iter_lines():
-                    if line:
-                        if line.strip() == b'data: [DONE]':
-                            break
-                        if line.startswith(b'data: '):
-                            try:
-                                chunk = json.loads(line.decode('utf-8').removeprefix('data: '))
-                                if content := chunk.get('choices', [{}])[0].get('delta', {}).get('content'):
-                                    yield content
-                            except json.JSONDecodeError:
-                                continue
-        except requests.RequestException as e:
-            raise LLMError(f"Stream request failed: {str(e)}")
-
-    def _send_request(self, payload: Dict[str, Any]) -> str:
-        """Send a non-streaming chat request.
-
-        Args:
-            payload: The prepared chat payload
-
-        Returns:
-            The complete response
-
-        Raises:
-            LLMError: If the request fails 🚫
-
-        Examples:
-            >>> llm = LLM("meta-llama/Meta-Llama-3-70B-Instruct")
-            >>> response = llm._send_request(llm._prepare_payload([
-            ...     {"role": "user", "content": "Tell me a joke!"}
-            ... ]))
-            >>> print(response)
-        """
-        try:
-            response = requests.post(self.api_url, json=payload, headers=self.headers)
-            response.raise_for_status()
-            result = response.json()
-            return result['choices'][0]['message']['content']
-        except requests.RequestException as e:
-            raise LLMError(f"Request failed: {str(e)}")
-        except (KeyError, IndexError) as e:
-            raise LLMError(f"Invalid response format: {str(e)}")
-        except json.JSONDecodeError as e:
-            raise LLMError(f"Invalid JSON response: {str(e)}")
-
-
-class VLM:
-    """Your gateway to vision-language AI magic! 🖼️
-
-    This class lets you:
-    - Chat about images with AI 🎨
-    - Get detailed image descriptions 📝
-    - Answer questions about images 🤔
-    - Stream responses in real-time ⚡
-
-    Examples:
-        >>> from webscout.LLM import VLM
-        >>> vlm = VLM("cogvlm-grounding-generalist")
-        >>> # Chat about an image
-        >>> response = vlm.chat([{
-        ...     "role": "user",
-        ...     "content": [
-        ...         {"type": "image", "image_url": "path/to/image.jpg"},
-        ...         {"type": "text", "text": "What's in this image?"}
-        ...     ]
-        ... }])
-        >>> print(response)
-        'I see a beautiful sunset over mountains...'
-    """
-
-    def __init__(self, model: str, system_message: str = "You are a Helpful AI."):
-        """Get ready for some vision-language magic! 🚀
-
-        Args:
-            model: Your chosen vision model
-            system_message: Set the AI's personality
-
-        Examples:
-            >>> vlm = VLM("cogvlm-grounding-generalist")
-            >>> print(vlm.model)
-            'cogvlm-grounding-generalist'
-        """
-        self.model = model
-        self.api_url = "https://api.deepinfra.com/v1/openai/chat/completions"
-        self.conversation_history = [{"role": "system", "content": system_message}]
-        self.headers = {
-            'User-Agent': 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/119.0.0.0 Safari/537.36',
-            'Accept-Language': 'en,fr-FR;q=0.9,fr;q=0.8,es-ES;q=0.7,es;q=0.6,en-US;q=0.5,am;q=0.4,de;q=0.3',
-            'Cache-Control': 'no-cache',
-            'Connection': 'keep-alive',
-            'Content-Type': 'application/json',
-            'Origin': 'https://deepinfra.com',
-            'Pragma': 'no-cache',
-            'Referer': 'https://deepinfra.com/',
-            'Sec-Fetch-Dest': 'empty',
-            'Sec-Fetch-Mode': 'cors',
-            'Sec-Fetch-Site': 'same-site',
-            'X-Deepinfra-Source': 'web-embed',
-            'accept': 'text/event-stream',
-            'sec-ch-ua': '"Google Chrome";v="119", "Chromium";v="119", "Not?A_Brand";v="24"',
-            'sec-ch-ua-mobile': '?0',
-            'sec-ch-ua-platform': '"macOS"'
-        }
-
-    def chat(
-        self, 
-        messages: List[Dict[str, Union[str, List[Dict[str, Union[str, Dict[str, str]]]]]]], 
-        stream: bool = False,
-        temperature: float = 0.7,
-        max_tokens: int = 8028,
-    ) -> Union[str, Generator[str, None, None]]:
-        """Chat about images with AI! 🖼️
-
-        This method lets you:
-        - Ask questions about images 🤔
-        - Get detailed descriptions 📝
-        - Stream responses in real-time ⚡
-        - Control response creativity 🎨
-
-        Args:
-            messages: Your chat + image data
-            stream: Want real-time responses?
-            temperature: Creativity level (0-1)
-            max_tokens: Max words to generate
-
-        Returns:
-            Either a complete response or streaming generator
-
-        Raises:
-            LLMError: If something goes wrong 🚫
-
-        Examples:
-            >>> vlm = VLM("cogvlm-grounding-generalist")
-            >>> # Regular chat with image
-            >>> response = vlm.chat([{
-            ...     "role": "user",
-            ...     "content": [
-            ...         {"type": "image", "image_url": "sunset.jpg"},
-            ...         {"type": "text", "text": "Describe this scene"}
-            ...     ]
-            ... }])
-            >>> # Streaming chat
-            >>> for chunk in vlm.chat([...], stream=True):
-            ...     print(chunk, end='')
-        """
-        payload = {
-            "model": self.model,
-            "messages": messages,
-            "stream": stream,
-            "temperature": temperature,
-            "max_tokens": max_tokens
-        }
-
-        try:
-            if stream:
-                return self._stream_response(payload)
-            else:
-                return self._send_request(payload)
-        except Exception as e:
-            raise LLMError(f"VLM API request failed: {str(e)}")
-
-    def _stream_response(self, payload: Dict[str, Any]) -> Generator[str, None, None]:
-        """Stream the VLM chat response."""
-        try:
-            with requests.post(self.api_url, json=payload, headers=self.headers, stream=True) as response:
-                response.raise_for_status()
-                for line in response.iter_lines():
-                    if line:
-                        if line.strip() == b'data: [DONE]':
-                            break
-                        if line.startswith(b'data: '):
-                            try:
-                                chunk = json.loads(line.decode('utf-8').removeprefix('data: '))
-                                if content := chunk.get('choices', [{}])[0].get('delta', {}).get('content'):
-                                    yield content
-                            except json.JSONDecodeError:
-                                continue
-        except requests.RequestException as e:
-            raise LLMError(f"VLM stream request failed: {str(e)}")
-
-    def _send_request(self, payload: Dict[str, Any]) -> str:
-        """Send a non-streaming VLM chat request."""
-        try:
-            response = requests.post(self.api_url, json=payload, headers=self.headers)
-            response.raise_for_status()
-            result = response.json()
-            return result['choices'][0]['message']['content']
-        except requests.RequestException as e:
-            raise LLMError(f"VLM request failed: {str(e)}")
-        except (KeyError, IndexError) as e:
-            raise LLMError(f"Invalid VLM response format: {str(e)}")
-        except json.JSONDecodeError as e:
-            raise LLMError(f"Invalid VLM JSON response: {str(e)}")
-
-
-def encode_image_to_base64(image_path: str) -> str:
-    """Turn your image into base64 magic! 🎨
-
-    Args:
-        image_path: Where's your image at?
-
-    Returns:
-        Your image as a base64 string ✨
-
-    Raises:
-        IOError: If we can't read your image 🚫
-
-    Examples:
-        >>> from webscout.LLM import encode_image_to_base64
-        >>> image_data = encode_image_to_base64("cool_pic.jpg")
-        >>> print(len(image_data))  # Check the encoded length
-        12345
-    """
-    try:
-        with open(image_path, "rb") as image_file:
-            return base64.b64encode(image_file.read()).decode("utf-8")
-    except IOError as e:
-        raise LLMError(f"Failed to read image file: {str(e)}")
-
-
-if __name__ == "__main__":
-    # Example usage
-    try:
-        # Initialize LLM with Llama 3 model
-        llm = LLM(model="Qwen/Qwen2.5-Coder-32B-Instruct")
-        
-        # Example messages
-        messages = [
-            {"role": "user", "content": "Write a short poem about AI."}
-        ]
-        
-        # Example 1: Non-streaming response
-        print("\nNon-streaming response:")
-        response = llm.chat(messages, stream=False)
-        print(response)
-        
-        # Example 2: Streaming response
-        print("\nStreaming response:")
-        for chunk in llm.chat(messages, stream=True):
-            print(chunk, end='', flush=True)
-        print("\n")
-        
-    except LLMError as e:
-        print(f"Error: {str(e)}")
-    except KeyboardInterrupt:
-        print("\nOperation cancelled by user")
+"""
+>>> from webscout.LLM import LLM, VLM
+>>> llm = LLM("meta-llama/Meta-Llama-3-70B-Instruct")
+>>> response = llm.chat([{"role": "user", "content": "What's good?"}])
+>>> print(response)
+'Hey! I'm doing great, thanks for asking! How can I help you today? 😊'
+
+>>> # For vision tasks
+>>> vlm = VLM("cogvlm-grounding-generalist")
+>>> response = vlm.chat([{"role": "user", "content": [{"type": "image", "image_url": "path/to/image.jpg"}, {"type": "text", "text": "What's in this image?"}]}])
+"""
+
+import requests
+import base64
+import json
+from typing import List, Dict, Union, Generator, Optional, Any
+
+class LLMError(Exception):
+    """Custom exception for LLM API errors 🚫
+
+    Examples:
+        >>> try:
+        ...     raise LLMError("API key not found!")
+        ... except LLMError as e:
+        ...     print(f"Error: {e}")
+        Error: API key not found!
+    """
+    pass
+
+class LLM:
+    """A class for chatting with DeepInfra's powerful language models! 🚀
+
+    This class lets you:
+    - Chat with state-of-the-art language models 💬
+    - Stream responses in real-time ⚡
+    - Control temperature and token limits 🎮
+    - Handle system messages and chat history 📝
+
+    Examples:
+        >>> from webscout.LLM import LLM
+        >>> llm = LLM("meta-llama/Meta-Llama-3-70B-Instruct")
+        >>> response = llm.chat([
+        ...     {"role": "user", "content": "Write a short poem!"}
+        ... ])
+        >>> print(response)
+        'Through starlit skies and morning dew,
+        Nature's beauty, forever new.
+        In every moment, magic gleams,
+        Life's poetry flows like gentle streams.'
+    """
+    
+    def __init__(self, model: str, system_message: str = "You are a Helpful AI."):
+        """
+        Initialize the LLM client.
+        
+        Args:
+            model: The model identifier (e.g., "meta-llama/Meta-Llama-3-70B-Instruct")
+            system_message: The system message to use for the conversation
+            
+        Examples:
+            >>> llm = LLM("meta-llama/Meta-Llama-3-70B-Instruct")
+            >>> print(llm.model)
+            'meta-llama/Meta-Llama-3-70B-Instruct'
+        """
+        self.model = model
+        self.api_url = "https://api.deepinfra.com/v1/openai/chat/completions"
+        self.conversation_history = [{"role": "system", "content": system_message}]
+        self.headers = {
+            'User-Agent': 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/119.0.0.0 Safari/537.36',
+            'Accept-Language': 'en,fr-FR;q=0.9,fr;q=0.8,es-ES;q=0.7,es;q=0.6,en-US;q=0.5,am;q=0.4,de;q=0.3',
+            'Cache-Control': 'no-cache',
+            'Connection': 'keep-alive',
+            'Content-Type': 'application/json',
+            'Origin': 'https://deepinfra.com',
+            'Pragma': 'no-cache',
+            'Referer': 'https://deepinfra.com/',
+            'Sec-Fetch-Dest': 'empty',
+            'Sec-Fetch-Mode': 'cors',
+            'Sec-Fetch-Site': 'same-site',
+            'X-Deepinfra-Source': 'web-embed',
+            'accept': 'text/event-stream',
+            'sec-ch-ua': '"Google Chrome";v="119", "Chromium";v="119", "Not?A_Brand";v="24"',
+            'sec-ch-ua-mobile': '?0',
+            'sec-ch-ua-platform': '"macOS"'
+        }
+
+    def _prepare_payload(
+        self, 
+        messages: List[Dict[str, str]], 
+        stream: bool = False,
+        temperature: float = 0.7,
+        max_tokens: int = 8028,
+        stop: Optional[List[str]] = None,
+    ) -> Dict[str, Any]:
+        """Prepare the chat payload with all the right settings! 🎯
+
+        Args:
+            messages: Your chat messages (role & content)
+            stream: Want real-time responses? Set True! ⚡
+            temperature: Creativity level (0-1) 🎨
+            max_tokens: Max words to generate 📝
+            stop: Words to stop at (optional) 🛑
+
+        Returns:
+            Dict with all the API settings ready to go! 🚀
+
+        Examples:
+            >>> llm = LLM("meta-llama/Meta-Llama-3-70B-Instruct")
+            >>> payload = llm._prepare_payload([
+            ...     {"role": "user", "content": "Hi!"}
+            ... ])
+            >>> print(payload['model'])
+            'meta-llama/Meta-Llama-3-70B-Instruct'
+        """
+        return {
+            'model': self.model,
+            'messages': messages,
+            'temperature': temperature,
+            'max_tokens': max_tokens,
+            'stop': stop or [],
+            'stream': stream
+        }
+
+    def chat(
+        self, 
+        messages: List[Dict[str, str]], 
+        stream: bool = False,
+        temperature: float = 0.7,
+        max_tokens: int = 8028,
+        stop: Optional[List[str]] = None,
+    ) -> Union[str, Generator[str, None, None]]:
+        """Start chatting with the AI! 💬
+
+        This method is your gateway to:
+        - Having awesome conversations 🗣️
+        - Getting creative responses 🎨
+        - Streaming real-time replies ⚡
+        - Controlling the output style 🎮
+
+        Args:
+            messages: Your chat messages (role & content)
+            stream: Want real-time responses? Set True!
+            temperature: Creativity level (0-1)
+            max_tokens: Max words to generate
+            stop: Words to stop at (optional)
+
+        Returns:
+            Either a complete response or streaming generator
+
+        Raises:
+            LLMError: If something goes wrong 🚫
+
+        Examples:
+            >>> llm = LLM("meta-llama/Meta-Llama-3-70B-Instruct")
+            >>> # Regular chat
+            >>> response = llm.chat([
+            ...     {"role": "user", "content": "Tell me a joke!"}
+            ... ])
+            >>> # Streaming chat
+            >>> for chunk in llm.chat([
+            ...     {"role": "user", "content": "Tell me a story!"}
+            ... ], stream=True):
+            ...     print(chunk, end='')
+        """
+        payload = self._prepare_payload(messages, stream, temperature, max_tokens, stop)
+        
+        try:
+            if stream:
+                return self._stream_response(payload)
+            else:
+                return self._send_request(payload)
+        except Exception as e:
+            raise LLMError(f"API request failed: {str(e)}")
+
+    def _stream_response(self, payload: Dict[str, Any]) -> Generator[str, None, None]:
+        """Stream the chat response in real-time! ⚡
+
+        Args:
+            payload: The prepared chat payload
+
+        Yields:
+            Streaming chunks of the response
+
+        Raises:
+            LLMError: If the stream request fails 🚫
+
+        Examples:
+            >>> llm = LLM("meta-llama/Meta-Llama-3-70B-Instruct")
+            >>> for chunk in llm._stream_response(llm._prepare_payload([
+            ...     {"role": "user", "content": "Tell me a story!"}
+            ... ])):
+            ...     print(chunk, end='')
+        """
+        try:
+            with requests.post(self.api_url, json=payload, headers=self.headers, stream=True) as response:
+                response.raise_for_status()
+                for line in response.iter_lines():
+                    if line:
+                        if line.strip() == b'data: [DONE]':
+                            break
+                        if line.startswith(b'data: '):
+                            try:
+                                chunk = json.loads(line.decode('utf-8').removeprefix('data: '))
+                                if content := chunk.get('choices', [{}])[0].get('delta', {}).get('content'):
+                                    yield content
+                            except json.JSONDecodeError:
+                                continue
+        except requests.RequestException as e:
+            raise LLMError(f"Stream request failed: {str(e)}")
+
+    def _send_request(self, payload: Dict[str, Any]) -> str:
+        """Send a non-streaming chat request.
+
+        Args:
+            payload: The prepared chat payload
+
+        Returns:
+            The complete response
+
+        Raises:
+            LLMError: If the request fails 🚫
+
+        Examples:
+            >>> llm = LLM("meta-llama/Meta-Llama-3-70B-Instruct")
+            >>> response = llm._send_request(llm._prepare_payload([
+            ...     {"role": "user", "content": "Tell me a joke!"}
+            ... ]))
+            >>> print(response)
+        """
+        try:
+            response = requests.post(self.api_url, json=payload, headers=self.headers)
+            response.raise_for_status()
+            result = response.json()
+            return result['choices'][0]['message']['content']
+        except requests.RequestException as e:
+            raise LLMError(f"Request failed: {str(e)}")
+        except (KeyError, IndexError) as e:
+            raise LLMError(f"Invalid response format: {str(e)}")
+        except json.JSONDecodeError as e:
+            raise LLMError(f"Invalid JSON response: {str(e)}")
+
+
+class VLM:
+    """Your gateway to vision-language AI magic! 🖼️
+
+    This class lets you:
+    - Chat about images with AI 🎨
+    - Get detailed image descriptions 📝
+    - Answer questions about images 🤔
+    - Stream responses in real-time ⚡
+
+    Examples:
+        >>> from webscout.LLM import VLM
+        >>> vlm = VLM("cogvlm-grounding-generalist")
+        >>> # Chat about an image
+        >>> response = vlm.chat([{
+        ...     "role": "user",
+        ...     "content": [
+        ...         {"type": "image", "image_url": "path/to/image.jpg"},
+        ...         {"type": "text", "text": "What's in this image?"}
+        ...     ]
+        ... }])
+        >>> print(response)
+        'I see a beautiful sunset over mountains...'
+    """
+
+    def __init__(self, model: str, system_message: str = "You are a Helpful AI."):
+        """Get ready for some vision-language magic! 🚀
+
+        Args:
+            model: Your chosen vision model
+            system_message: Set the AI's personality
+
+        Examples:
+            >>> vlm = VLM("cogvlm-grounding-generalist")
+            >>> print(vlm.model)
+            'cogvlm-grounding-generalist'
+        """
+        self.model = model
+        self.api_url = "https://api.deepinfra.com/v1/openai/chat/completions"
+        self.conversation_history = [{"role": "system", "content": system_message}]
+        self.headers = {
+            'User-Agent': 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/119.0.0.0 Safari/537.36',
+            'Accept-Language': 'en,fr-FR;q=0.9,fr;q=0.8,es-ES;q=0.7,es;q=0.6,en-US;q=0.5,am;q=0.4,de;q=0.3',
+            'Cache-Control': 'no-cache',
+            'Connection': 'keep-alive',
+            'Content-Type': 'application/json',
+            'Origin': 'https://deepinfra.com',
+            'Pragma': 'no-cache',
+            'Referer': 'https://deepinfra.com/',
+            'Sec-Fetch-Dest': 'empty',
+            'Sec-Fetch-Mode': 'cors',
+            'Sec-Fetch-Site': 'same-site',
+            'X-Deepinfra-Source': 'web-embed',
+            'accept': 'text/event-stream',
+            'sec-ch-ua': '"Google Chrome";v="119", "Chromium";v="119", "Not?A_Brand";v="24"',
+            'sec-ch-ua-mobile': '?0',
+            'sec-ch-ua-platform': '"macOS"'
+        }
+
+    def chat(
+        self, 
+        messages: List[Dict[str, Union[str, List[Dict[str, Union[str, Dict[str, str]]]]]]], 
+        stream: bool = False,
+        temperature: float = 0.7,
+        max_tokens: int = 8028,
+    ) -> Union[str, Generator[str, None, None]]:
+        """Chat about images with AI! 🖼️
+
+        This method lets you:
+        - Ask questions about images 🤔
+        - Get detailed descriptions 📝
+        - Stream responses in real-time ⚡
+        - Control response creativity 🎨
+
+        Args:
+            messages: Your chat + image data
+            stream: Want real-time responses?
+            temperature: Creativity level (0-1)
+            max_tokens: Max words to generate
+
+        Returns:
+            Either a complete response or streaming generator
+
+        Raises:
+            LLMError: If something goes wrong 🚫
+
+        Examples:
+            >>> vlm = VLM("cogvlm-grounding-generalist")
+            >>> # Regular chat with image
+            >>> response = vlm.chat([{
+            ...     "role": "user",
+            ...     "content": [
+            ...         {"type": "image", "image_url": "sunset.jpg"},
+            ...         {"type": "text", "text": "Describe this scene"}
+            ...     ]
+            ... }])
+            >>> # Streaming chat
+            >>> for chunk in vlm.chat([...], stream=True):
+            ...     print(chunk, end='')
+        """
+        payload = {
+            "model": self.model,
+            "messages": messages,
+            "stream": stream,
+            "temperature": temperature,
+            "max_tokens": max_tokens
+        }
+
+        try:
+            if stream:
+                return self._stream_response(payload)
+            else:
+                return self._send_request(payload)
+        except Exception as e:
+            raise LLMError(f"VLM API request failed: {str(e)}")
+
+    def _stream_response(self, payload: Dict[str, Any]) -> Generator[str, None, None]:
+        """Stream the VLM chat response."""
+        try:
+            with requests.post(self.api_url, json=payload, headers=self.headers, stream=True) as response:
+                response.raise_for_status()
+                for line in response.iter_lines():
+                    if line:
+                        if line.strip() == b'data: [DONE]':
+                            break
+                        if line.startswith(b'data: '):
+                            try:
+                                chunk = json.loads(line.decode('utf-8').removeprefix('data: '))
+                                if content := chunk.get('choices', [{}])[0].get('delta', {}).get('content'):
+                                    yield content
+                            except json.JSONDecodeError:
+                                continue
+        except requests.RequestException as e:
+            raise LLMError(f"VLM stream request failed: {str(e)}")
+
+    def _send_request(self, payload: Dict[str, Any]) -> str:
+        """Send a non-streaming VLM chat request."""
+        try:
+            response = requests.post(self.api_url, json=payload, headers=self.headers)
+            response.raise_for_status()
+            result = response.json()
+            return result['choices'][0]['message']['content']
+        except requests.RequestException as e:
+            raise LLMError(f"VLM request failed: {str(e)}")
+        except (KeyError, IndexError) as e:
+            raise LLMError(f"Invalid VLM response format: {str(e)}")
+        except json.JSONDecodeError as e:
+            raise LLMError(f"Invalid VLM JSON response: {str(e)}")
+
+
+def encode_image_to_base64(image_path: str) -> str:
+    """Turn your image into base64 magic! 🎨
+
+    Args:
+        image_path: Where's your image at?
+
+    Returns:
+        Your image as a base64 string ✨
+
+    Raises:
+        IOError: If we can't read your image 🚫
+
+    Examples:
+        >>> from webscout.LLM import encode_image_to_base64
+        >>> image_data = encode_image_to_base64("cool_pic.jpg")
+        >>> print(len(image_data))  # Check the encoded length
+        12345
+    """
+    try:
+        with open(image_path, "rb") as image_file:
+            return base64.b64encode(image_file.read()).decode("utf-8")
+    except IOError as e:
+        raise LLMError(f"Failed to read image file: {str(e)}")
+
+
+if __name__ == "__main__":
+    # Example usage
+    try:
+        # Initialize LLM with Llama 3 model
+        llm = LLM(model="mistralai/Mistral-Small-24B-Instruct-2501")
+        
+        # Example messages
+        messages = [
+            {"role": "user", "content": "Write a short poem about AI."}
+        ]
+        
+        # Example 1: Non-streaming response
+        print("\nNon-streaming response:")
+        response = llm.chat(messages, stream=False)
+        print(response)
+        
+        # Example 2: Streaming response
+        print("\nStreaming response:")
+        for chunk in llm.chat(messages, stream=True):
+            print(chunk, end='', flush=True)
+        print("\n")
+        
+    except LLMError as e:
+        print(f"Error: {str(e)}")
+    except KeyboardInterrupt:
+        print("\nOperation cancelled by user")