{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "6511a91c-ed20-41ff-befb-699bda1912a3",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-03-25T08:44:53.813781Z",
     "iopub.status.busy": "2026-03-25T08:44:53.813650Z",
     "iopub.status.idle": "2026-03-25T08:45:08.246826Z",
     "shell.execute_reply": "2026-03-25T08:45:08.246118Z",
     "shell.execute_reply.started": "2026-03-25T08:44:53.813766Z"
    }
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "61572014a82e4ee0a3756b44f486678f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Downloading (incomplete total...): 0.00B [00:00, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "75e4fe58bc664443a44cfa0287da5a71",
       "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": "707b86cddd9147eeada08aab3314962f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Loading weights:   0%|          | 0/824 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[text] batch=0, tokens=4, pos=0..3 (t=h=w): [0, 1, 2, 3]\n",
      "get_vision_position_ids: grid_thw=tensor([ 1, 18, 18], device='cuda:0'), llm_grid_thw=(1, 9, 9), start_position=4\n",
      "  temp_merge_size=1, spatial_merge_size=2\n",
      "  image_seq_length=81\n",
      "  position_width (repeat)=[4, 5, 6, 7, 8, 9, 10, 11, 12, 4]...[12, 4, 5, 6, 7, 8, 9, 10, 11, 12]\n",
      "  position_height (repeat_interleave)=[4, 4, 4, 4, 4, 4, 4, 4, 4, 5]...[11, 12, 12, 12, 12, 12, 12, 12, 12, 12]\n",
      "  position_temporal (torch.full) (before spacing)=[4, 4, 4, 4, 4, 4, 4, 4, 4, 4]...[4, 4, 4, 4, 4, 4, 4, 4, 4, 4]\n",
      "  time_interval=2\n",
      "  position_temporal (after spacing)=[8, 8, 8, 8, 8, 8, 8, 8, 8, 8]...[8, 8, 8, 8, 8, 8, 8, 8, 8, 8]\n",
      "[vision pos] grid_thw=tensor([ 1, 18, 18], device='cuda:0'), start=4\n",
      "  t: [8, 8, 8, 8, 8, 8, 8, 8, 8, 8]...[8, 8, 8, 8, 8, 8, 8, 8, 8, 8]\n",
      "  h: [4, 4, 4, 4, 4, 4, 4, 4, 4, 5]...[11, 12, 12, 12, 12, 12, 12, 12, 12, 12]\n",
      "  w: [4, 5, 6, 7, 8, 9, 10, 11, 12, 4]...[12, 4, 5, 6, 7, 8, 9, 10, 11, 12]\n",
      "[text] batch=0, tokens=9, pos=13..21 (t=h=w): [13, 14, 15, 16, 17, 18, 19, 20, 21]\n",
      "[LLM prefill] position_ids shape: torch.Size([3, 1, 94])  (3=t/h/w, bs, seq_len)\n",
      "  batch 0 (shape: 94):\n",
      "    t: [0, 1, 2, 3, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 13, 14, 15, 16, 17, 18, 19, 20, 21] \n",
      "    h: [0, 1, 2, 3, 4, 4, 4, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 5, 5, 5, 5, 5, 6, 6, 6, 6, 6, 6, 6, 6, 6, 7, 7, 7, 7, 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 8, 8, 8, 9, 9, 9, 9, 9, 9, 9, 9, 9, 10, 10, 10, 10, 10, 10, 10, 10, 10, 11, 11, 11, 11, 11, 11, 11, 11, 11, 12, 12, 12, 12, 12, 12, 12, 12, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21] \n",
      "    w: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 4, 5, 6, 7, 8, 9, 10, 11, 12, 4, 5, 6, 7, 8, 9, 10, 11, 12, 4, 5, 6, 7, 8, 9, 10, 11, 12, 4, 5, 6, 7, 8, 9, 10, 11, 12, 4, 5, 6, 7, 8, 9, 10, 11, 12, 4, 5, 6, 7, 8, 9, 10, 11, 12, 4, 5, 6, 7, 8, 9, 10, 11, 12, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21] \n",
      "\n",
      "Manual Prompt Response: The image shows a black Porsche Panamera parked on a road. The car is positioned at an angle, with the rear of the vehicle facing the camera. The Porsche logo and model name are visible on the trunk. The car has a sleek design with a long hood and short rear deck. The taillights are prominent, and the license plate is clearly visible. The background appears to be a blurred landscape, suggesting motion or a focus on the car itself.\n"
     ]
    }
   ],
   "source": [
    "import torch\n",
    "from transformers import Qwen2_5_VLForConditionalGeneration, AutoProcessor\n",
    "from qwen_vl_utils import process_vision_info\n",
    "\n",
    "# 1. Load Model and Processor\n",
    "model_name = \"Qwen/Qwen2.5-VL-3B-Instruct\"\n",
    "model = Qwen2_5_VLForConditionalGeneration.from_pretrained(\n",
    "    model_name, torch_dtype=torch.float16, device_map='auto'\n",
    ")\n",
    "processor = AutoProcessor.from_pretrained(model_name)\n",
    "\n",
    "# 2. Define your inputs manually\n",
    "image_url = \"./car-1_256_0.jpg\"\n",
    "user_query = \"Describe the image\"\n",
    "\n",
    "# 3. Construct the prompt string manually\n",
    "# Qwen2.5-VL expects specific tokens to wrap system, user, and assistant roles.\n",
    "# Note: The <|vision_start|> and <|vision_end|> tags tell the processor \n",
    "# where to inject the image features.\n",
    "prompt = (\n",
    "    \"<|im_start|>user\\n\"\n",
    "    \"<|vision_start|><|image_pad|><|vision_end|>\"\n",
    "    f\"{user_query}<|im_end|>\\n\"\n",
    "    \"<|im_start|>assistant\\n\"\n",
    ")\n",
    "\n",
    "# 4. Process the vision information\n",
    "# We still use this utility to fetch the image and handle resizing logic\n",
    "messages = [{\"role\": \"user\", \"content\": [{\"type\": \"image\", \"image\": image_url}]}]\n",
    "image_inputs, _ = process_vision_info(messages)\n",
    "\n",
    "# 5. Tokenize and Prepare Tensors\n",
    "inputs = processor(\n",
    "    text=[prompt],\n",
    "    images=image_inputs,\n",
    "    videos=None,\n",
    "    padding=True,\n",
    "    return_tensors=\"pt\",\n",
    ")\n",
    "inputs = inputs.to(model.device)\n",
    "\n",
    "# 6. Generate\n",
    "generated_ids = model.generate(**inputs, max_new_tokens=100)\n",
    "\n",
    "# Trim the prompt tokens from the result\n",
    "generated_ids_trimmed = [\n",
    "    out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)\n",
    "]\n",
    "\n",
    "output_text = processor.batch_decode(\n",
    "    generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False\n",
    ")\n",
    "\n",
    "print(f\"\\nManual Prompt Response: {output_text[0]}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "f45df021-6302-4f47-9e06-8070577885a2",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-03-25T08:45:08.247563Z",
     "iopub.status.busy": "2026-03-25T08:45:08.247274Z",
     "iopub.status.idle": "2026-03-25T08:45:08.251422Z",
     "shell.execute_reply": "2026-03-25T08:45:08.250728Z",
     "shell.execute_reply.started": "2026-03-25T08:45:08.247548Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'<|im_start|>user\\n<|vision_start|><|image_pad|><|vision_end|>Describe the image<|im_end|>\\n<|im_start|>assistant\\n'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "prompt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "504fa71b-42b4-4f53-8988-25fcfba38d13",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-03-25T08:45:08.252004Z",
     "iopub.status.busy": "2026-03-25T08:45:08.251839Z",
     "iopub.status.idle": "2026-03-25T08:45:08.262159Z",
     "shell.execute_reply": "2026-03-25T08:45:08.261665Z",
     "shell.execute_reply.started": "2026-03-25T08:45:08.251990Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(torch.Size([1, 1, 94, 128]), torch.float16)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cos = torch.load('cos.pt')\n",
    "cos.shape, cos.dtype"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "a874f8d4-efa9-4ba8-9e57-c019da0775bd",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-03-25T08:45:08.262852Z",
     "iopub.status.busy": "2026-03-25T08:45:08.262705Z",
     "iopub.status.idle": "2026-03-25T08:45:08.269483Z",
     "shell.execute_reply": "2026-03-25T08:45:08.268795Z",
     "shell.execute_reply.started": "2026-03-25T08:45:08.262840Z"
    }
   },
   "outputs": [],
   "source": [
    "torch.set_printoptions(precision=14)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "f44460e3-58e9-4fd2-898a-06e8a00f9365",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-03-25T08:45:08.269891Z",
     "iopub.status.busy": "2026-03-25T08:45:08.269762Z",
     "iopub.status.idle": "2026-03-25T08:45:08.295575Z",
     "shell.execute_reply": "2026-03-25T08:45:08.294849Z",
     "shell.execute_reply.started": "2026-03-25T08:45:08.269879Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[[0.54052734375000, 0.69238281250000, 0.79638671875000,\n",
       "          0.86621093750000, 0.91259765625000, 0.94287109375000,\n",
       "          0.96289062500000, 0.97558593750000, 0.98437500000000,\n",
       "          0.98974609375000, 0.99316406250000, 0.99560546875000,\n",
       "          0.99707031250000, 0.99804687500000, 0.99902343750000,\n",
       "          0.99902343750000, 0.99951171875000, 0.99951171875000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 0.54052734375000, 0.69238281250000,\n",
       "          0.79638671875000, 0.86621093750000, 0.91259765625000,\n",
       "          0.94287109375000, 0.96289062500000, 0.97558593750000,\n",
       "          0.98437500000000, 0.98974609375000, 0.99316406250000,\n",
       "          0.99560546875000, 0.99707031250000, 0.99804687500000,\n",
       "          0.99902343750000, 0.99902343750000, 0.99951171875000,\n",
       "          0.99951171875000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000, 1.00000000000000,\n",
       "          1.00000000000000, 1.00000000000000]]], device='cuda:0',\n",
       "       dtype=torch.float16)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cos[:, :, 1, :]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "039f1456-ffa4-40b2-8ba1-a0cd5f74733e",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-03-25T08:45:08.296250Z",
     "iopub.status.busy": "2026-03-25T08:45:08.296111Z",
     "iopub.status.idle": "2026-03-25T08:45:08.348040Z",
     "shell.execute_reply": "2026-03-25T08:45:08.347477Z",
     "shell.execute_reply.started": "2026-03-25T08:45:08.296237Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 torch.Size([128])\n",
      "   [0:16] 1.0 1.0\n",
      "   [16:40] 1.0 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] 1.0 1.0\n",
      "   [80:104] 1.0 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "1 torch.Size([128])\n",
      "   [0:16] 0.54052734375 0.9990234375\n",
      "   [16:40] 0.99951171875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] 0.54052734375 0.9990234375\n",
      "   [80:104] 0.99951171875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "2 torch.Size([128])\n",
      "   [0:16] -0.416259765625 0.9970703125\n",
      "   [16:40] 0.998046875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.416259765625 0.9970703125\n",
      "   [80:104] 0.998046875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "3 torch.Size([128])\n",
      "   [0:16] -0.990234375 0.9931640625\n",
      "   [16:40] 0.99560546875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.990234375 0.9931640625\n",
      "   [80:104] 0.99560546875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "4 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9921875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9921875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "5 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9921875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9921875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "6 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9921875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9921875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "7 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9921875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9921875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "8 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9921875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9921875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "9 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9921875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9921875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "10 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9921875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9921875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "11 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9921875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9921875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "12 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9921875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9921875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "13 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9873046875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9873046875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "14 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9873046875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9873046875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "15 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9873046875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9873046875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "16 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9873046875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9873046875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "17 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9873046875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9873046875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "18 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9873046875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9873046875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "19 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9873046875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9873046875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "20 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9873046875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9873046875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "21 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9873046875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9873046875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "22 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.98193359375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.98193359375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "23 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.98193359375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.98193359375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "24 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.98193359375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.98193359375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "25 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.98193359375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.98193359375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "26 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.98193359375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.98193359375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "27 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.98193359375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.98193359375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "28 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.98193359375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.98193359375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "29 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.98193359375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.98193359375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "30 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.98193359375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.98193359375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "31 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9755859375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9755859375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "32 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9755859375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9755859375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "33 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9755859375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9755859375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "34 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9755859375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9755859375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "35 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9755859375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9755859375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "36 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9755859375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9755859375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "37 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9755859375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9755859375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "38 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9755859375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9755859375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "39 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9755859375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9755859375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "40 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.96826171875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.96826171875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "41 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.96826171875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.96826171875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "42 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.96826171875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.96826171875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "43 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.96826171875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.96826171875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "44 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.96826171875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.96826171875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "45 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.96826171875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.96826171875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "46 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.96826171875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.96826171875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "47 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.96826171875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.96826171875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "48 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.96826171875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.96826171875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "49 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9599609375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9599609375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "50 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9599609375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9599609375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "51 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9599609375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9599609375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "52 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9599609375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9599609375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "53 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9599609375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9599609375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "54 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9599609375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9599609375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "55 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9599609375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9599609375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "56 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9599609375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9599609375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "57 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9599609375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9599609375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "58 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9501953125 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9501953125 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "59 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9501953125 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9501953125 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "60 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9501953125 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9501953125 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "61 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9501953125 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9501953125 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "62 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9501953125 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9501953125 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "63 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9501953125 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9501953125 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "64 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9501953125 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9501953125 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "65 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9501953125 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9501953125 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "66 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9501953125 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9501953125 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "67 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.93994140625 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.93994140625 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "68 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.93994140625 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.93994140625 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "69 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.93994140625 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.93994140625 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "70 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.93994140625 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.93994140625 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "71 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.93994140625 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.93994140625 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "72 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.93994140625 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.93994140625 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "73 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.93994140625 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.93994140625 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "74 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.93994140625 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.93994140625 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "75 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.93994140625 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.93994140625 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "76 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9287109375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9287109375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "77 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9287109375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9287109375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "78 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9287109375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9287109375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "79 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9287109375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9287109375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "80 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9287109375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9287109375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "81 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9287109375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9287109375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "82 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9287109375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9287109375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "83 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9287109375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9287109375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "84 torch.Size([128])\n",
      "   [0:16] -0.97314453125 0.98681640625\n",
      "   [16:40] 0.9287109375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97314453125 0.98681640625\n",
      "   [80:104] 0.9287109375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "85 torch.Size([128])\n",
      "   [0:16] -0.962890625 0.9072265625\n",
      "   [16:40] 0.91650390625 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.962890625 0.9072265625\n",
      "   [80:104] 0.91650390625 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "86 torch.Size([128])\n",
      "   [0:16] -0.99853515625 0.9287109375\n",
      "   [16:40] 0.90380859375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.99853515625 0.9287109375\n",
      "   [80:104] 0.90380859375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "87 torch.Size([128])\n",
      "   [0:16] -0.98583984375 0.9990234375\n",
      "   [16:40] 0.8896484375 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.98583984375 0.9990234375\n",
      "   [80:104] 0.8896484375 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "88 torch.Size([128])\n",
      "   [0:16] -0.95751953125 0.94677734375\n",
      "   [16:40] 0.87451171875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.95751953125 0.94677734375\n",
      "   [80:104] 0.87451171875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "89 torch.Size([128])\n",
      "   [0:16] -0.9931640625 0.87451171875\n",
      "   [16:40] 0.85888671875 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.9931640625 0.87451171875\n",
      "   [80:104] 0.85888671875 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "90 torch.Size([128])\n",
      "   [0:16] -1.0 0.986328125\n",
      "   [16:40] 0.84228515625 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -1.0 0.986328125\n",
      "   [80:104] 0.84228515625 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "91 torch.Size([128])\n",
      "   [0:16] -0.97216796875 0.98876953125\n",
      "   [16:40] 0.82470703125 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.97216796875 0.98876953125\n",
      "   [80:104] 0.82470703125 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "92 torch.Size([128])\n",
      "   [0:16] -0.96240234375 0.91259765625\n",
      "   [16:40] 0.806640625 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.96240234375 0.91259765625\n",
      "   [80:104] 0.806640625 1.0\n",
      "   [104:128] 1.0 1.0\n",
      "93 torch.Size([128])\n",
      "   [0:16] -0.9912109375 0.861328125\n",
      "   [16:40] 0.78759765625 1.0\n",
      "   [40:64] 1.0 1.0\n",
      "   [64:80] -0.9912109375 0.861328125\n",
      "   [80:104] 0.78759765625 1.0\n",
      "   [104:128] 1.0 1.0\n"
     ]
    }
   ],
   "source": [
    "cos_reshaped = cos.squeeze(0).squeeze(0)\n",
    "for idx, tensor in enumerate(cos_reshaped):\n",
    "    print(idx, tensor.shape)\n",
    "    # print(tensor[16:40].flatten().tolist())\n",
    "    if tensor.shape[0] != 128:\n",
    "        print(f\"Skipping tensor {idx} with shape {tensor.shape}\")\n",
    "        continue\n",
    "    else:\n",
    "        print(f\"   [0:16] {tensor[0:16].min()} {tensor[0:16].max()}\")\n",
    "        print(f\"   [16:40] {tensor[16:40].min()} {tensor[16:40].max()}\")\n",
    "        print(f\"   [40:64] {tensor[40:64].min()} {tensor[40:64].max()}\")\n",
    "        print(f\"   [64:80] {tensor[64:80].min()} {tensor[64:80].max()}\")\n",
    "        print(f\"   [80:104] {tensor[80:104].min()} {tensor[80:104].max()}\")\n",
    "        print(f\"   [104:128] {tensor[104:128].min()} {tensor[104:128].max()}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e3745573-5a06-4beb-894a-7a9dddf82d6d",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "91f8dc55-fd0e-4d6c-84d1-626b8fae96a0",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-03-25T08:58:50.174004Z",
     "iopub.status.busy": "2026-03-25T08:58:50.173602Z",
     "iopub.status.idle": "2026-03-25T08:58:50.184512Z",
     "shell.execute_reply": "2026-03-25T08:58:50.183786Z",
     "shell.execute_reply.started": "2026-03-25T08:58:50.173987Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Python:       3.12.12 | packaged by Anaconda, Inc. | (main, Oct 21 2025, 20:16:04) [GCC 11.2.0]\n",
      "Torch:        2.10.0+cu128\n",
      "Transformers: 5.3.0.dev0\n",
      "Pillow:       12.1.1\n",
      "Torchvision:  0.25.0+cu128\n",
      "qwen_vl_utils: NOT INSTALLED or no __version__\n",
      "\n",
      "=== Processor Image Config ===\n",
      "size (min/max pixels): SizeDict(height=None, width=None, longest_edge=12845056, shortest_edge=3136, max_height=None, max_width=None)\n",
      "patch_size: 14\n",
      "merge_size: 2\n",
      "resample: 3\n",
      "\n",
      "=== Raw Image ===\n",
      "PIL size (WxH): (256, 256)\n",
      "PIL mode: RGB\n",
      "Pixel checksum (sum of all pixels): 18508962\n",
      "Pixel hash (first 10 rows md5):\n",
      "  51edee767ff6c404d737e620d51036d5\n",
      "\n",
      "=== Processor Outputs ===\n",
      "input_ids shape: torch.Size([1, 94])\n",
      "input_ids hash: d3e7a59feedd64667ccc03bd71448894\n",
      "pixel_values shape: torch.Size([324, 1176])\n",
      "pixel_values dtype: torch.float32\n",
      "pixel_values sum: -114413.40625\n",
      "pixel_values hash: a635309793003d45969e76ca7badaca0\n",
      "image_grid_thw: tensor([[ 1, 18, 18]], device='cuda:0')\n"
     ]
    }
   ],
   "source": [
    "## === REPRODUCIBILITY FINGERPRINT ===\n",
    "\n",
    "import sys, torch, transformers, PIL, torchvision\n",
    "print(\"Python:      \", sys.version)\n",
    "print(\"Torch:       \", torch.__version__)\n",
    "print(\"Transformers:\", transformers.__version__)\n",
    "print(\"Pillow:      \", PIL.__version__)\n",
    "print(\"Torchvision: \", torchvision.__version__)\n",
    "try:\n",
    "    import qwen_vl_utils\n",
    "    print(\"qwen_vl_utils:\", qwen_vl_utils.__version__)\n",
    "except Exception:\n",
    "    print(\"qwen_vl_utils: NOT INSTALLED or no __version__\")\n",
    "\n",
    "print()\n",
    "\n",
    "# --- Check processor config ---\n",
    "print(\"=== Processor Image Config ===\")\n",
    "print(\"size (min/max pixels):\", processor.image_processor.size)\n",
    "print(\"patch_size:\", processor.image_processor.patch_size)\n",
    "print(\"merge_size:\", processor.image_processor.merge_size)\n",
    "print(\"resample:\", processor.image_processor.resample)\n",
    "\n",
    "print()\n",
    "\n",
    "# --- Check image after loading (before processor) ---\n",
    "from PIL import Image\n",
    "img = Image.open(\"./car-1_256_0.jpg\").convert(\"RGB\")\n",
    "print(\"=== Raw Image ===\")\n",
    "print(\"PIL size (WxH):\", img.size)\n",
    "print(\"PIL mode:\", img.mode)\n",
    "\n",
    "import numpy as np\n",
    "img_np = np.array(img)\n",
    "print(\"Pixel checksum (sum of all pixels):\", img_np.sum())\n",
    "print(\"Pixel hash (first 10 rows md5):\")\n",
    "import hashlib\n",
    "print(\" \", hashlib.md5(img_np[:10].tobytes()).hexdigest())\n",
    "\n",
    "print()\n",
    "\n",
    "# --- Check inputs after processor ---\n",
    "print(\"=== Processor Outputs ===\")\n",
    "print(\"input_ids shape:\", inputs.input_ids.shape)\n",
    "print(\"input_ids hash:\", hashlib.md5(inputs.input_ids.cpu().numpy().tobytes()).hexdigest())\n",
    "print(\"pixel_values shape:\", inputs.pixel_values.shape)\n",
    "print(\"pixel_values dtype:\", inputs.pixel_values.dtype)\n",
    "print(\"pixel_values sum:\", inputs.pixel_values.sum().item())\n",
    "print(\"pixel_values hash:\", hashlib.md5(inputs.pixel_values.float().cpu().numpy().tobytes()).hexdigest())\n",
    "print(\"image_grid_thw:\", inputs.image_grid_thw)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d9e47144-a9f2-465f-9c4b-7458220e3510",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}