File size: 6,874 Bytes
571ff03 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 | from __future__ import annotations
import os
import base64
import mimetypes
import argparse
from pathlib import Path
from typing import List, Dict, Any, Optional
from dotenv import load_dotenv
from huggingface_hub import InferenceClient
# Load environment variables (e.g., HF_TOKEN) from a .env file
load_dotenv()
# This is the system prompt you provided
AGENT_SYSTEM_PROMPT = """You are a general AI assistant. I will ask you a question. Report your thoughts, and finish your answer with the following template: FINAL ANSWER: [YOUR FINAL ANSWER]. YOUR FINAL ANSWER should be a number OR as few words as possible OR a comma separated list of numbers and/or strings. If you are asked for a number, don't use comma to write your number neither use units such as $ or percent sign unless specified otherwise. If you are asked for a string, don't use articles, neither abbreviations (e.g. for cities), and write the digits in plain text unless specified otherwise. If you are asked for a comma separated list, apply the above rules depending of whether the element to be put in the list is a number or a string."""
class ImageAgent:
"""
Encapsulated agent for answering questions about an image.
Usage:
# Assumes HF_TOKEN is in your environment
agent = ImageAgent()
# Ask a question with an image
question = "What is in this image?"
image_path = "/path/to/your/image.png"
answer = agent.answer(question, image_path)
# Or use the callable shortcut
answer = agent(question, image_path)
"""
def __init__(
self,
model_id: Optional[str] = None,
provider: Optional[str] = "novita",
temperature: float = 0.2,
api_key: Optional[str] = None,
):
"""
Initializes the ImageAgent.
Args:
model_id: The repository ID of the model to use.
provider: The provider for the InferenceClient.
temperature: The generation temperature.
api_key: The API key. Defaults to os.environ["HF_TOKEN"].
"""
self.model_id = model_id or "meta-llama/Llama-4-Scout-17B-16E-Instruct"
self.temperature = temperature
# Get API key from arg or environment, raising an error if missing
self.api_key = api_key or os.environ.get("HF_TOKEN")
if not self.api_key:
raise ValueError(
"API key not found. Please set the HF_TOKEN environment variable "
"or pass it as 'api_key' to the ImageAgent."
)
# Initialize the client
self.client = InferenceClient(
provider=provider,
api_key=self.api_key,
)
# ---- Public API ----
def answer(self, question: str, image_path: str) -> str:
"""
Run the agent to answer a question about an image.
Args:
question: The text prompt to ask the model.
image_path: The local file path to the image.
Returns:
The parsed "Final Answer" string from the model.
"""
# 1. Convert image to data URL
try:
image_data_url = self._to_data_url(image_path)
except FileNotFoundError:
return f"Error: Image file not found at {image_path}"
except Exception as e:
return f"Error processing image: {e}"
# 2. Build the messages payload
messages = self._build_messages(question, image_data_url)
# 3. Call the API
try:
completion = self.client.chat.completions.create(
model=self.model_id,
messages=messages,
temperature=self.temperature,
)
response_text = completion.choices[0].message.content
except Exception as e:
return f"Error calling API: {e}"
# 4. Parse and return the final answer
return self._parse_answer(response_text)
def __call__(self, question: str, image_path: str) -> str:
"""Callable shortcut for self.answer()."""
return self.answer(question, image_path)
# ---- Internals ----
def _to_data_url(self, path: str) -> str:
"""Converts a local file path to a base64 data URL."""
p = Path(path)
if not p.exists():
raise FileNotFoundError(f"No file found at {path}")
mime = mimetypes.guess_type(str(p))[0] or "image/png"
b64 = base64.b64encode(p.read_bytes()).decode("utf-8")
return f"data:{mime};base64,{b64}"
def _build_messages(self, question: str, image_data_url: str) -> List[Dict[str, Any]]:
"""Constructs the message payload for the API."""
return [
{
"role": "system",
"content": [
{"type": "text", "text": AGENT_SYSTEM_PROMPT},
],
},
{
"role": "user",
"content": [
{"type": "text", "text": question},
{
"type": "image_url",
"image_url": {"url": image_data_url},
},
],
},
]
def _parse_answer(self, response: str) -> str:
"""Extracts the text after 'FINAL ANSWER: '."""
try:
# Use rpartition for robustness, splits on the *last* occurrence
_before, _marker, final_answer = response.rpartition("FINAL ANSWER: ")
if _marker:
return final_answer.strip()
else:
# If marker isn't found, return the whole response with a warning
print("Warning: Could not find 'FINAL ANSWER:' marker. Returning full response.")
return response.strip()
except Exception as e:
print(f"Error parsing response: {e}")
return response.strip()
# ---- CLI Entrypoint ----
def main():
"""
Allows running the agent from the command line.
Example:
python image_agent.py "What move is this?" "path/to/my_image.png"
"""
parser = argparse.ArgumentParser(description="Class-based agent for image questions.")
parser.add_argument("question", type=str, help="Your question (quoted).")
parser.add_argument("image_path", type=str, help="Path to the image file.")
args = parser.parse_args()
try:
image_agent = ImageAgent()
answer = image_agent.answer(args.question, args.image_path)
print(answer)
except Exception as e:
print(f"An error occurred: {e}")
if __name__ == "__main__":
main() |