{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import AutoProcessor, Gemma3ForConditionalGeneration, TextIteratorStreamer\n",
    "import torch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_id =  \"google/gemma-3-12b-it\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "\n",
    "\n",
    "processor = AutoProcessor.from_pretrained(model_id, padding_side=\"left\")\n",
    "model = Gemma3ForConditionalGeneration.from_pretrained(\n",
    "    model_id, device_map=\"auto\", torch_dtype=torch.bfloat16, attn_implementation=\"eager\"\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "97908a20851e455c895464dc7228dbb6",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Fetching 2 files:   0%|          | 0/2 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c8011adeae5247f89612a93c3ff9e963",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "model-00001-of-00002.safetensors:   2%|1         | 83.9M/4.96G [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b604d0bd3be24ddda117cc36a8752c8e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "model-00002-of-00002.safetensors:   3%|3         | 126M/3.64G [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Error while downloading from https://cdn-lfs-us-1.hf.co/repos/83/76/8376859a3a783fbbf8c6b8aff73e386e0379657f480bf946d481f9a936d4ceab/fdde0e5aa5ced0fa203b3d50f4ab78168b7e3a3e08c6349f5cc9326666e1bb13?response-content-disposition=inline%3B+filename*%3DUTF-8%27%27model-00002-of-00002.safetensors%3B+filename%3D%22model-00002-of-00002.safetensors%22%3B&Expires=1742502952&Policy=eyJTdGF0ZW1lbnQiOlt7IkNvbmRpdGlvbiI6eyJEYXRlTGVzc1RoYW4iOnsiQVdTOkVwb2NoVGltZSI6MTc0MjUwMjk1Mn19LCJSZXNvdXJjZSI6Imh0dHBzOi8vY2RuLWxmcy11cy0xLmhmLmNvL3JlcG9zLzgzLzc2LzgzNzY4NTlhM2E3ODNmYmJmOGM2YjhhZmY3M2UzODZlMDM3OTY1N2Y0ODBiZjk0NmQ0ODFmOWE5MzZkNGNlYWIvZmRkZTBlNWFhNWNlZDBmYTIwM2IzZDUwZjRhYjc4MTY4YjdlM2EzZTA4YzYzNDlmNWNjOTMyNjY2NmUxYmIxMz9yZXNwb25zZS1jb250ZW50LWRpc3Bvc2l0aW9uPSoifV19&Signature=h9yYRgT8Ui5-lDdTma5cMxNLJQn3a5qkQb1B9ICLWd2hPe68qnwhgkSyyze48E9LkP%7EstZozqs-6jQFBLFnbShc098FQdwXXw4NonedJMgex2loENeJ4Juaw7JVdhyUo1J1k8BM0j3t97kO%7ESsL2XN7mJ-DdqQqr6bbD5KmD3kxwaEMOt0Bo9EXSt%7EPUxFV1v9Z91m2XrJjjmgLKZEOY05K8p15-3VVhIFlQe1PcGVb72lpp3nMaXgPFoMJVY6NUVKIhszzJDFc6JhBfeO31Rb6Q9YmO-mm8duFSRCxUhqYzVMx-E4MN0iJO3lMrK0aTDXm6Ps8Fk1v8Rk4Bj6auew__&Key-Pair-Id=K24J24Z295AEI9: HTTPSConnectionPool(host='cdn-lfs-us-1.hf.co', port=443): Read timed out.\n",
      "Trying to resume download...\n",
      "Error while downloading from https://cdn-lfs-us-1.hf.co/repos/83/76/8376859a3a783fbbf8c6b8aff73e386e0379657f480bf946d481f9a936d4ceab/eb5fd5e97ddd07b56778733e9653c07312529cb00980a318fc3e1c4e3b5a8f1f?response-content-disposition=inline%3B+filename*%3DUTF-8%27%27model-00001-of-00002.safetensors%3B+filename%3D%22model-00001-of-00002.safetensors%22%3B&Expires=1742502952&Policy=eyJTdGF0ZW1lbnQiOlt7IkNvbmRpdGlvbiI6eyJEYXRlTGVzc1RoYW4iOnsiQVdTOkVwb2NoVGltZSI6MTc0MjUwMjk1Mn19LCJSZXNvdXJjZSI6Imh0dHBzOi8vY2RuLWxmcy11cy0xLmhmLmNvL3JlcG9zLzgzLzc2LzgzNzY4NTlhM2E3ODNmYmJmOGM2YjhhZmY3M2UzODZlMDM3OTY1N2Y0ODBiZjk0NmQ0ODFmOWE5MzZkNGNlYWIvZWI1ZmQ1ZTk3ZGRkMDdiNTY3Nzg3MzNlOTY1M2MwNzMxMjUyOWNiMDA5ODBhMzE4ZmMzZTFjNGUzYjVhOGYxZj9yZXNwb25zZS1jb250ZW50LWRpc3Bvc2l0aW9uPSoifV19&Signature=Vr9B3wdQIVtGgOocVA7IUl3gNH6pmGYUaeZ4TnqxQz7uO3BAXFQDsIcffFn1hePw-huNkRXwKBLMiPc7wD%7E6nAi4INXD7ydOfb8NDcIuzk0TZfHm0d4RheWz14OWuD4gohXh1QAa%7EF7dxn0X2OvOJ1C8hPlElE6G8g9NtGYncfWCASAkjsmFhyGMDg8i5MAt5hU5oscdJ0gpeckc1OjRNdqEISslG2jHSiTBeEn18UIyLMB3OGSxuaiguPoGPFpBUCTIkeSkUFbWCnnwOLkO7gD6hLCVw9F%7EiBZN4r92VdBlWCeRb6A%7ELH46Vo480H742XAACEEdXrqHSI2VWq9xAA__&Key-Pair-Id=K24J24Z295AEI9: HTTPSConnectionPool(host='cdn-lfs-us-1.hf.co', port=443): Read timed out.\n",
      "Trying to resume download...\n"
     ]
    }
   ],
   "source": [
    "from transformers import pipeline\n",
    "import torch\n",
    "\n",
    "pipe = pipeline(\n",
    "    \"image-text-to-text\",\n",
    "    model=\"google/gemma-3-4b-it\",\n",
    "    device=\"cuda\",\n",
    "    torch_dtype=torch.bfloat16,\n",
    "    cache_dir=\"F:\\\\huggingface_cache\"\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# pip install accelerate\n",
    "print(\"Hi\")\n",
    "from transformers import AutoProcessor, Gemma3ForConditionalGeneration\n",
    "import requests\n",
    "import torch\n",
    "from PIL import Image\n",
    "\n",
    "print(\"Done\")\n",
    "model_id = \"google/gemma-3-4b-it\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "pip install bitsandbytes\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "from transformers import Gemma3ForConditionalGeneration, AutoProcessor\n",
    "from transformers import BitsAndBytesConfig\n",
    "device = \"cuda\" if torch.cuda.is_available() else \"cpu\"\n",
    "print(f\"Using device: {device}\")\n",
    "\n",
    "model_id = \"google/gemma-3-4b-it\"\n",
    "\n",
    "\n",
    "quantization_config = BitsAndBytesConfig(load_in_8bit=True)\n",
    "# Load the model and move it to the correct device\n",
    "model = Gemma3ForConditionalGeneration.from_pretrained(\n",
    "    model_id,\n",
    "    cache_dir=\"F:\\\\huggingface_cache\",\n",
    "    device_map=\"auto\",  # Automatically assigns layers to available devices\n",
    "    quantization_config=quantization_config\n",
    ").eval()\n",
    "\n",
    "# Load the processor\n",
    "processor = AutoProcessor.from_pretrained(model_id, cache_dir=\"F:\\\\huggingface_cache\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "messages = [\n",
    "    {\n",
    "        \"role\": \"system\",\n",
    "        \"content\": [{\"type\": \"text\", \"text\": \"You are a helpful assistant.\"}]\n",
    "    },\n",
    "    {\n",
    "        \"role\": \"user\",\n",
    "        \"content\": [\n",
    "            \n",
    "            {\"type\": \"text\", \"text\": \"Whats the color of sky?.\"}\n",
    "        ]\n",
    "    }\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "inputs = processor.apply_chat_template(\n",
    "    messages, add_generation_prompt=True, tokenize=True,\n",
    "    return_dict=True, return_tensors=\"pt\"\n",
    ").to(model.device, dtype=torch.bfloat16)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "input_len = inputs[\"input_ids\"].shape[-1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "with torch.inference_mode():\n",
    "    generation = model.generate(**inputs, max_new_tokens=100, do_sample=False)\n",
    "    generation = generation[0][input_len:]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "with torch.inference_mode():\n",
    "    generation = model.generate(**inputs, max_new_tokens=100, do_sample=False)\n",
    "    generation = generation[0][input_len:]\n",
    "\n",
    "decoded = processor.decode(generation, skip_special_tokens=True)\n",
    "print(decoded)\n",
    "\n",
    "# **Overall Impression:** The image is a close-up shot of a vibrant garden scene, \n",
    "# focusing on a cluster of pink cosmos flowers and a busy bumblebee. \n",
    "# It has a slightly soft, natural feel, likely captured in daylight.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "access_token='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'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Use a pipeline as a high-level helper\n",
    "from transformers import pipeline\n",
    "\n",
    "messages = [\n",
    "    {\"role\": \"user\", \"content\": \"Who are you?\"},\n",
    "]\n",
    "pipe = pipeline(\"image-text-to-text\", model=\"google/gemma-3-4b-it\")\n",
    "pipe(messages)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "hf_env_kernel",
   "language": "python",
   "name": "hf_env"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}