{ "cells": [ { "cell_type": "code", "execution_count": null, "metadata": { "id": "xL8y37Y6bORU" }, "outputs": [], "source": [ "%%capture\n", "!pip install gradio spaces transformers accelerate numpy requests\n", "!pip install torch torchvision qwen-vl-utils av hf_xet\n", "!pip install pillow huggingface_hub opencv-python" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "id": "Y-NTbL1tdL9X" }, "outputs": [], "source": [ "import os\n", "import random\n", "import uuid\n", "import json\n", "import time\n", "import asyncio\n", "from threading import Thread\n", "\n", "import gradio as gr\n", "import spaces\n", "import torch\n", "import numpy as np\n", "from PIL import Image\n", "import cv2\n", "\n", "from transformers import (\n", " Qwen2VLForConditionalGeneration,\n", " AutoProcessor,\n", " TextIteratorStreamer,\n", ")\n", "from transformers.image_utils import load_image\n", "\n", "# Constants for text generation\n", "MAX_MAX_NEW_TOKENS = 2048\n", "DEFAULT_MAX_NEW_TOKENS = 1024\n", "# Increase or disable input truncation to avoid token mismatches\n", "MAX_INPUT_TOKEN_LENGTH = int(os.getenv(\"MAX_INPUT_TOKEN_LENGTH\", \"8192\"))\n", "\n", "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n", "\n", "MODEL_ID = \"Qwen/Qwen2-VL-2B-Instruct\"\n", "processor = AutoProcessor.from_pretrained(MODEL_ID, trust_remote_code=True)\n", "model_m = Qwen2VLForConditionalGeneration.from_pretrained(\n", " MODEL_ID,\n", " trust_remote_code=True,\n", " torch_dtype=torch.float16\n", ").to(\"cuda\").eval()\n", "\n", "def downsample_video(video_path):\n", " \"\"\"\n", " Downsamples the video to evenly spaced frames.\n", " Each frame is returned as a PIL image along with its timestamp.\n", " \"\"\"\n", " vidcap = cv2.VideoCapture(video_path)\n", " total_frames = int(vidcap.get(cv2.CAP_PROP_FRAME_COUNT))\n", " fps = vidcap.get(cv2.CAP_PROP_FPS)\n", " frames = []\n", " # Sample 10 evenly spaced frames.\n", " frame_indices = np.linspace(0, total_frames - 1, 10, dtype=int)\n", " for i in frame_indices:\n", " vidcap.set(cv2.CAP_PROP_POS_FRAMES, i)\n", " success, image = vidcap.read()\n", " if success:\n", " image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB) # Convert BGR to RGB\n", " pil_image = Image.fromarray(image)\n", " timestamp = round(i / fps, 2)\n", " frames.append((pil_image, timestamp))\n", " vidcap.release()\n", " return frames\n", "\n", "@spaces.GPU\n", "def generate_image(text: str, image: Image.Image,\n", " max_new_tokens: int = 1024,\n", " temperature: float = 0.6,\n", " top_p: float = 0.9,\n", " top_k: int = 50,\n", " repetition_penalty: float = 1.2):\n", "\n", " if image is None:\n", " yield \"Please upload an image.\"\n", " return\n", "\n", " messages = [{\n", " \"role\": \"user\",\n", " \"content\": [\n", " {\"type\": \"image\", \"image\": image},\n", " {\"type\": \"text\", \"text\": text},\n", " ]\n", " }]\n", " prompt_full = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)\n", " inputs = processor(\n", " text=[prompt_full],\n", " images=[image],\n", " return_tensors=\"pt\",\n", " padding=True,\n", " truncation=False,\n", " max_length=MAX_INPUT_TOKEN_LENGTH\n", " ).to(\"cuda\")\n", " streamer = TextIteratorStreamer(processor, skip_prompt=True, skip_special_tokens=True)\n", " generation_kwargs = {**inputs, \"streamer\": streamer, \"max_new_tokens\": max_new_tokens}\n", " thread = Thread(target=model_m.generate, kwargs=generation_kwargs)\n", " thread.start()\n", " buffer = \"\"\n", " for new_text in streamer:\n", " buffer += new_text\n", " buffer = buffer.replace(\"<|im_end|>\", \"\")\n", " time.sleep(0.01)\n", " yield buffer\n", "\n", "@spaces.GPU\n", "def generate_video(text: str, video_path: str,\n", " max_new_tokens: int = 1024,\n", " temperature: float = 0.6,\n", " top_p: float = 0.9,\n", " top_k: int = 50,\n", " repetition_penalty: float = 1.2):\n", "\n", " if video_path is None:\n", " yield \"Please upload a video.\"\n", " return\n", "\n", " frames = downsample_video(video_path)\n", " messages = [\n", " {\"role\": \"system\", \"content\": [{\"type\": \"text\", \"text\": \"You are a helpful assistant.\"}]},\n", " {\"role\": \"user\", \"content\": [{\"type\": \"text\", \"text\": text}]}\n", " ]\n", " # Append each frame with its timestamp.\n", " for frame in frames:\n", " image, timestamp = frame\n", " messages[1][\"content\"].append({\"type\": \"text\", \"text\": f\"Frame {timestamp}:\"})\n", " messages[1][\"content\"].append({\"type\": \"image\", \"image\": image})\n", " inputs = processor.apply_chat_template(\n", " messages,\n", " tokenize=True,\n", " add_generation_prompt=True,\n", " return_dict=True,\n", " return_tensors=\"pt\",\n", " truncation=False,\n", " max_length=MAX_INPUT_TOKEN_LENGTH\n", " ).to(\"cuda\")\n", " streamer = TextIteratorStreamer(processor, skip_prompt=True, skip_special_tokens=True)\n", " generation_kwargs = {\n", " **inputs,\n", " \"streamer\": streamer,\n", " \"max_new_tokens\": max_new_tokens,\n", " \"do_sample\": True,\n", " \"temperature\": temperature,\n", " \"top_p\": top_p,\n", " \"top_k\": top_k,\n", " \"repetition_penalty\": repetition_penalty,\n", " }\n", " thread = Thread(target=model_m.generate, kwargs=generation_kwargs)\n", " thread.start()\n", " buffer = \"\"\n", " for new_text in streamer:\n", " buffer += new_text\n", " buffer = buffer.replace(\"<|im_end|>\", \"\")\n", " time.sleep(0.01)\n", " yield buffer\n", "\n", "css = \"\"\"\n", ".submit-btn {\n", " background-color: #2980b9 !important;\n", " color: white !important;\n", "}\n", ".submit-btn:hover {\n", " background-color: #3498db !important;\n", "}\n", "\"\"\"\n", "\n", "# Create the Gradio Interface\n", "with gr.Blocks(css=css, theme=\"bethecloud/storj_theme\") as demo:\n", " gr.Markdown(\"# **Qwen/Qwen2-VL-2B-Instruct**\")\n", " with gr.Row():\n", " with gr.Column():\n", " with gr.Tabs():\n", " with gr.TabItem(\"Image Inference\"):\n", " image_query = gr.Textbox(label=\"Query Input\", placeholder=\"Enter your query here...\")\n", " image_upload = gr.Image(type=\"pil\", label=\"Image\")\n", " image_submit = gr.Button(\"Submit\", elem_classes=\"submit-btn\")\n", "\n", " with gr.TabItem(\"Video Inference\"):\n", " video_query = gr.Textbox(label=\"Query Input\", placeholder=\"Enter your query here...\")\n", " video_upload = gr.Video(label=\"Video\")\n", " video_submit = gr.Button(\"Submit\", elem_classes=\"submit-btn\")\n", "\n", " with gr.Accordion(\"Advanced options\", open=False):\n", " max_new_tokens = gr.Slider(label=\"Max new tokens\", minimum=1, maximum=MAX_MAX_NEW_TOKENS, step=1, value=DEFAULT_MAX_NEW_TOKENS)\n", " temperature = gr.Slider(label=\"Temperature\", minimum=0.1, maximum=4.0, step=0.1, value=0.6)\n", " top_p = gr.Slider(label=\"Top-p (nucleus sampling)\", minimum=0.05, maximum=1.0, step=0.05, value=0.9)\n", " top_k = gr.Slider(label=\"Top-k\", minimum=1, maximum=1000, step=1, value=50)\n", " repetition_penalty = gr.Slider(label=\"Repetition penalty\", minimum=1.0, maximum=2.0, step=0.05, value=1.2)\n", " with gr.Column():\n", " output = gr.Textbox(label=\"Output\", interactive=False)\n", "\n", " image_submit.click(\n", " fn=generate_image,\n", " inputs=[image_query, image_upload, max_new_tokens, temperature, top_p, top_k, repetition_penalty],\n", " outputs=output\n", " )\n", " video_submit.click(\n", " fn=generate_video,\n", " inputs=[video_query, video_upload, max_new_tokens, temperature, top_p, top_k, repetition_penalty],\n", " outputs=output\n", " )\n", "\n", "if __name__ == \"__main__\":\n", " demo.queue(max_size=30).launch(share=True, ssr_mode=False, show_error=True)" ] } ], "metadata": { "accelerator": "GPU", "colab": { "gpuType": "T4", "provenance": [] }, "kernelspec": { "display_name": "Python 3", "name": "python3" }, "language_info": { "name": "python" } }, "nbformat": 4, "nbformat_minor": 0 }