Large Language Models for Text Generation.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "21cdcef8",
   "metadata": {},
   "source": [
    "# HW1- LLM- Bahareh Golchin"
   ]
  },
  {
   "cell_type": "raw",
   "id": "01f4b25c",
   "metadata": {},
   "source": [
    "# Q1. Differences Between Llama 3.2 Models and Phi-3.5 Model"
   ]
  },
  {
   "cell_type": "raw",
   "id": "0c8ceb2d",
   "metadata": {},
   "source": [
    "1-Model Architecture and Size:\n",
    "\n",
    "Llama 3.2 Models (1B and 3B parameters) are based on a transformer architecture that is optimized for various tasks, including text generation. The smaller model (1B) is less capable than the larger model (3B) but is faster and requires less memory.\n",
    "Phi-3.5 Model is also based on a transformer architecture but is specifically designed for instruction-based tasks, which means it may incorporate additional features for better understanding and following user instructions.\n",
    "\n",
    "2-Training Objectives:\n",
    "\n",
    "Llama Models are generally trained with a focus on generating coherent and contextually relevant text but do not specifically emphasize instruction-following capabilities.\n",
    "Phi-3.5 Model, as indicated by its name, likely incorporates training techniques aimed at enhancing its ability to understand and execute explicit user instructions, making it more suitable for conversational AI applications.\n",
    "\n",
    "3-Performance and Use Cases:\n",
    "\n",
    "Llama 3.2 Models may excel in creative text generation and storytelling applications but might struggle with more complex task-oriented requests.\n",
    "Phi-3.5 Model is designed for instruction-based tasks, making it more effective in scenarios where following user commands and producing relevant responses is critical."
   ]
  },
  {
   "cell_type": "raw",
   "id": "211bf133",
   "metadata": {},
   "source": [
    "Q2 & Q3"
   ]
  },
  {
   "cell_type": "raw",
   "id": "a7201d3c",
   "metadata": {},
   "source": [
    "Llama-3.2-1B"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "db264ef7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5ded854f176b4897a80bc3dbb5262c16",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "VBox(children=(HTML(value='<center> <img\\nsrc=https://huggingface.co/front/assets/huggingface_logo-noborder.sv…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Note: Environment variable`HF_TOKEN` is set and is the current active token independently from the token you've just configured.\n"
     ]
    }
   ],
   "source": [
    "#Setup\n",
    "import os\n",
    "import torch\n",
    "from huggingface_hub import login\n",
    "from transformers import AutoTokenizer, AutoModelForCausalLM\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "\n",
    "hf_token = os.environ.get('TOKEN')\n",
    "\n",
    "# Login to Hugging Face using the secret token\n",
    "login(token=hf_token)\n",
    "\n",
    "# Load Model\n",
    "models = {\n",
    "    'Llama-3.2-1B': {\n",
    "        'tokenizer': AutoTokenizer.from_pretrained(\"meta-llama/Llama-3.2-1B\"),\n",
    "        'model': AutoModelForCausalLM.from_pretrained(\"meta-llama/Llama-3.2-1B\")\n",
    "    }\n",
    "}\n",
    "\n",
    "# Assign the tokenizer and model to variables for use\n",
    "tokenizer = models['Llama-3.2-1B']['tokenizer']\n",
    "model = models['Llama-3.2-1B']['model']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "583dbd11",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "The attention mask and the pad token id were not set. As a consequence, you may observe unexpected behavior. Please pass your input's `attention_mask` to obtain reliable results.\n",
      "Setting `pad_token_id` to `eos_token_id`:None for open-end generation.\n",
      "The attention mask is not set and cannot be inferred from input because pad token is same as eos token. As a consequence, you may observe unexpected behavior. Please pass your input's `attention_mask` to obtain reliable results.\n",
      "Starting from v4.46, the `logits` model output will have the same type as the model (except at train time, where it will always be FP32)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generated Story:\n",
      " Once upon a time there was a girl who lived in a small village in a small country called Ireland. She was very pretty and had a very sweet and gentle nature. She had a husband who was very kind and loving and they had two children. One of the children was a boy and the other was a girl.\n",
      "One day the girl's husband went out to work in the fields. She was very sad and worried about him and worried that he would not be back soon. She began to think about what she would do if he did not come home. She began to worry about what would happen to her and her children if something happened to her husband.\n",
      "One day she heard a knock on the door. She opened the door and saw a man standing there. She was very surprised to see him. He was a stranger and he did not look like anyone she knew. She said to him, \"Who are you?\" He said, \"I am your husband's brother. I have come\n",
      "\n",
      "Metrics for the generated story:\n",
      "   Type-Token Ratio  Perplexity\n",
      "0          0.564246    2.665452\n"
     ]
    }
   ],
   "source": [
    "# Generate Stories\n",
    "def generate_story(prompt=\"Once upon a time\", max_length=200):\n",
    "    inputs = tokenizer(prompt, return_tensors='pt')\n",
    "    outputs = model.generate(inputs['input_ids'], max_length=max_length)\n",
    "    story = tokenizer.decode(outputs[0], skip_special_tokens=True)\n",
    "    return story\n",
    "\n",
    "story = generate_story()\n",
    "print(\"Generated Story:\\n\", story)\n",
    "\n",
    "# Metrics Calculation\n",
    "def compute_metrics(text):\n",
    "    tokens = text.split()\n",
    "    type_token_ratio = len(set(tokens)) / len(tokens) if len(tokens) > 0 else 0\n",
    "    perplexity = np.exp(-np.mean([np.log((len(tokens) - i) / len(tokens)) for i in range(len(tokens))]))\n",
    "    return type_token_ratio, perplexity\n",
    "\n",
    "metrics = compute_metrics(story)\n",
    "metrics_df = pd.DataFrame([metrics], columns=['Type-Token Ratio', 'Perplexity'])\n",
    "print(\"\\nMetrics for the generated story:\")\n",
    "print(metrics_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "3b9bd2ed",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Model Size\n",
    "model_sizes = {'Llama-3.2-1B': 1e9}  # 1 billion parameters\n",
    "sizes = list(model_sizes.values())\n",
    "ttr = metrics_df['Type-Token Ratio']\n",
    "perplexity = metrics_df['Perplexity']\n",
    "\n",
    "# Plotting TTR\n",
    "plt.figure(figsize=(10, 5))\n",
    "plt.scatter(sizes, ttr)\n",
    "plt.title('Model Size vs. Type-Token Ratio')\n",
    "plt.xlabel('Model Size (Parameters)')\n",
    "plt.ylabel('Type-Token Ratio')\n",
    "plt.grid(True)\n",
    "plt.xticks(sizes, ['Llama-3.2-1B'], rotation=45)\n",
    "plt.show()\n",
    "\n",
    "# Plotting Perplexity\n",
    "plt.figure(figsize=(10, 5))\n",
    "plt.scatter(sizes, perplexity)\n",
    "plt.title('Model Size vs. Perplexity')\n",
    "plt.xlabel('Model Size (Parameters)')\n",
    "plt.ylabel('Perplexity')\n",
    "plt.grid(True)\n",
    "plt.xticks(sizes, ['Llama-3.2-1B'], rotation=45)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "raw",
   "id": "3d58c1de",
   "metadata": {},
   "source": [
    "Llama-3.2-3B"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "51691348",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d5d1d7ccaf0d43ad8dbd9e876bb8aa54",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Loading checkpoint shards:   0%|          | 0/2 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "models = {\n",
    "    'Llama-3.2-3B': {\n",
    "        'tokenizer': AutoTokenizer.from_pretrained(\"meta-llama/Llama-3.2-3B\"),\n",
    "        'model': AutoModelForCausalLM.from_pretrained(\"meta-llama/Llama-3.2-3B\")\n",
    "    }\n",
    "}\n",
    "\n",
    "# Assign the tokenizer and model to variables for use\n",
    "tokenizer = models['Llama-3.2-3B']['tokenizer']\n",
    "model = models['Llama-3.2-3B']['model']\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "bbc5c558",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "The attention mask and the pad token id were not set. As a consequence, you may observe unexpected behavior. Please pass your input's `attention_mask` to obtain reliable results.\n",
      "Setting `pad_token_id` to `eos_token_id`:None for open-end generation.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generated Story:\n",
      " Once upon a time, in a land far, far away, there was a boy who lived in a small village. The boy was named Jack and he loved to play with his friends. One day, while playing with his friends, Jack stumbled upon a magic wand. The wand had a strange power – when Jack waved it, a magic portal opened up, and Jack found himself in a new world. This new world was full of wonders and adventures. Jack was thrilled to explore this new world, and he made many friends along the way. One of his best friends was a girl named Jill. Jill was a skilled archer and a brave warrior. She taught Jack how to shoot arrows and fight monsters. Together, they faced many challenges and triumphed over them. But one day, Jill and Jack found themselves in a dark and dangerous place. They were surrounded by monsters and they were running out of time. Jack and Jill were in a tight spot, but they knew they had to work\n",
      "\n",
      "Metrics for the generated story:\n",
      "   Type-Token Ratio  Perplexity\n",
      "0          0.614943    2.664166\n"
     ]
    }
   ],
   "source": [
    "#Generate Stories\n",
    "def generate_story(prompt=\"Once upon a time\", max_length=200):\n",
    "    inputs = tokenizer(prompt, return_tensors='pt')\n",
    "    outputs = model.generate(inputs['input_ids'], max_length=max_length)\n",
    "    story = tokenizer.decode(outputs[0], skip_special_tokens=True)\n",
    "    return story\n",
    "\n",
    "# Generate a story\n",
    "story = generate_story()\n",
    "print(\"Generated Story:\\n\", story)\n",
    "\n",
    "# Metrics Calculation\n",
    "def compute_metrics(text):\n",
    "    tokens = text.split()\n",
    "    type_token_ratio = len(set(tokens)) / len(tokens) if len(tokens) > 0 else 0\n",
    "    perplexity = np.exp(-np.mean([np.log((len(tokens) - i) / len(tokens)) for i in range(len(tokens))]))  # Ensure this line is correctly formatted\n",
    "\n",
    "    return type_token_ratio, perplexity\n",
    "\n",
    "# Calculate metrics for the generated story\n",
    "metrics = compute_metrics(story)\n",
    "metrics_df = pd.DataFrame([metrics], columns=['Type-Token Ratio', 'Perplexity'])\n",
    "\n",
    "# Display Metrics\n",
    "print(\"\\nMetrics for the generated story:\")\n",
    "print(metrics_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "9f26668b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model_sizes = {'Llama-3.2-3B': 3e9}  # 3 billion parameters\n",
    "sizes = list(model_sizes.values())\n",
    "ttr = metrics_df['Type-Token Ratio']\n",
    "perplexity = metrics_df['Perplexity']\n",
    "\n",
    "# Plotting TTR\n",
    "plt.figure(figsize=(10, 5))\n",
    "plt.scatter(sizes, ttr)\n",
    "plt.title('Model Size vs. Type-Token Ratio')\n",
    "plt.xlabel('Model Size (Parameters)')\n",
    "plt.ylabel('Type-Token Ratio')\n",
    "plt.grid(True)\n",
    "plt.xticks(sizes, ['Llama-3.2-3B'], rotation=45)\n",
    "plt.show()\n",
    "\n",
    "# Plotting Perplexity\n",
    "plt.figure(figsize=(10, 5))\n",
    "plt.scatter(sizes, perplexity)\n",
    "plt.title('Model Size vs. Perplexity')\n",
    "plt.xlabel('Model Size (Parameters)')\n",
    "plt.ylabel('Perplexity')\n",
    "plt.grid(True)\n",
    "plt.xticks(sizes, ['Llama-3.2-3B'], rotation=45)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "raw",
   "id": "195fcdaa",
   "metadata": {},
   "source": [
    "Phi-3.5-mini-instruct"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "ab5bfa0e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "094a710d156640e38335befa63b575eb",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Loading checkpoint shards:   0%|          | 0/2 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Load the models and tokenizers\n",
    "models = {\n",
    "    'Phi-3.5-mini-instruct': {\n",
    "        'tokenizer': AutoTokenizer.from_pretrained(\"microsoft/Phi-3.5-mini-instruct\"),\n",
    "        'model': AutoModelForCausalLM.from_pretrained(\"microsoft/Phi-3.5-mini-instruct\")\n",
    "    }\n",
    "}\n",
    "\n",
    "# Assign the tokenizer and model to variables for use\n",
    "tokenizer = models['Phi-3.5-mini-instruct']['tokenizer']\n",
    "model = models['Phi-3.5-mini-instruct']['model']\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "d042e2e3",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Starting from v4.46, the `logits` model output will have the same type as the model (except at train time, where it will always be FP32)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generated Story:\n",
      " Once upon a time, in a small town nestled between rolling hills and lush greenery, there lived a young girl named Lily. Lily was a curious and adventurous soul, always seeking new experiences and knowledge. She had a deep love for nature and spent most of her days exploring the nearby forests and meadows.\n",
      "\n",
      "One sunny morning, as Lily ventured deeper into the woods, she stumbled upon a hidden clearing. In the center of the clearing stood a magnificent oak tree, its branches stretching towards the sky like outstretched arms. Intrigued, Lily approached the tree and noticed a small, weathered sign hanging from one of its branches. It read, \"The Enchanted Garden: A Haven of Wonders.\"\n",
      "\n",
      "Driven by her insatiable curiosity, Lily followed the sign's instructions and found herself standing at the entrance of a magical garden\n",
      "\n",
      "Metrics for the generated story:\n",
      "   Type-Token Ratio  Perplexity\n",
      "0          0.773723    2.652047\n"
     ]
    }
   ],
   "source": [
    "# Generate Stories\n",
    "def generate_story(prompt=\"Once upon a time\", max_length=200):\n",
    "    inputs = tokenizer(prompt, return_tensors='pt')\n",
    "    outputs = model.generate(inputs['input_ids'], max_length=max_length)\n",
    "    story = tokenizer.decode(outputs[0], skip_special_tokens=True)\n",
    "    return story\n",
    "\n",
    "# Generate a story\n",
    "story = generate_story()\n",
    "print(\"Generated Story:\\n\", story)\n",
    "\n",
    "# Metrics Calculation\n",
    "def compute_metrics(text):\n",
    "    tokens = text.split()\n",
    "    type_token_ratio = len(set(tokens)) / len(tokens) if len(tokens) > 0 else 0\n",
    "    perplexity = np.exp(-np.mean([np.log((len(tokens) - i) / len(tokens)) for i in range(len(tokens))]))  # Ensure this line is correctly formatted\n",
    "\n",
    "    return type_token_ratio, perplexity\n",
    "\n",
    "# Calculate metrics for the generated story\n",
    "metrics = compute_metrics(story)\n",
    "metrics_df = pd.DataFrame([metrics], columns=['Type-Token Ratio', 'Perplexity'])\n",
    "\n",
    "# Display Metrics\n",
    "print(\"\\nMetrics for the generated story:\")\n",
    "print(metrics_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "e8fbd563",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model_sizes = {'Phi-3.5-mini-instruct': 1e8}  # 100 million parameters\n",
    "sizes = list(model_sizes.values())\n",
    "ttr = metrics_df['Type-Token Ratio']\n",
    "perplexity = metrics_df['Perplexity']\n",
    "\n",
    "# Plotting TTR\n",
    "plt.figure(figsize=(10, 5))\n",
    "plt.scatter(sizes, ttr)\n",
    "plt.title('Model Size vs. Type-Token Ratio')\n",
    "plt.xlabel('Model Size (Parameters)')\n",
    "plt.ylabel('Type-Token Ratio')\n",
    "plt.grid(True)\n",
    "plt.xticks(sizes, ['Phi-3.5-mini-instruct'], rotation=45)\n",
    "plt.show()\n",
    "\n",
    "# Plotting Perplexity\n",
    "plt.figure(figsize=(10, 5))\n",
    "plt.scatter(sizes, perplexity)\n",
    "plt.title('Model Size vs. Perplexity')\n",
    "plt.xlabel('Model Size (Parameters)')\n",
    "plt.ylabel('Perplexity')\n",
    "plt.grid(True)\n",
    "plt.xticks(sizes, ['Phi-3.5-mini-instruct'], rotation=45)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8c4abdef",
   "metadata": {},
   "source": [
    "Q3:\n",
    "\n",
    "The TTR increases as we move from the Llama-3.2-1B model to the Llama-3.2-3B model and reaches its highest value with the Phi-3.5-mini-instruct model.\n",
    "This indicates that larger models (especially Phi-3.5) produce more diverse vocabulary and creative outputs. The higher the TTR, the more varied the language used in the generated text.\n",
    "\n",
    "The Perplexity values remain relatively stable across the models, with a slight decrease as we progress from the Llama models to the Phi model. Lower perplexity values suggest improved coherence in the generated text.\n",
    "\n",
    "As model sizes increase, there is a clear benefit in terms of generating diverse language (as indicated by TTR) and maintaining coherence (as indicated by Perplexity). However, larger models also require more computational resources and longer processing times, which can limit their usability in real-time applications or environments with constrained resources."
   ]
  },
  {
   "cell_type": "raw",
   "id": "46a23e43",
   "metadata": {},
   "source": [
    "Q4:\n",
    "\n",
    "The Phi-3.5 model produced a more structured and coherent story compared to the Llama models. For example, the generated story has a clear narrative flow, starting with an introduction of the character (Lily) and her adventurous spirit, followed by a specific setting (the hidden clearing with the oak tree). This clarity makes the story engaging and easy to follow.\n",
    "\n",
    "While I used the same code to generate stories for all models, the Phi-3.5 model is designed to follow instructions better, which means it naturally produced a more organized story.\n",
    "The prompt \"Once upon a time\" is a classic storytelling start, and it works well with the Phi-3.5 model, helping it create a vivid narrative.\n",
    "Although I didn’t change the prompt, the model understood how to generate a good story from that simple start."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b13a24a4",
   "metadata": {},
   "source": [
    "Q5: \n",
    "Since Phi-3.5-mini-instruct is the best performing model based on my results, here are the steps and code for experimenting with various parameters."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "a86b5187",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "22f065458f6e4176be4d0542d831865e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "VBox(children=(HTML(value='<center> <img\\nsrc=https://huggingface.co/front/assets/huggingface_logo-noborder.sv…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b85e7451a5934c19be2a5d5183b333fd",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Loading checkpoint shards:   0%|          | 0/2 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\robotics\\AppData\\Roaming\\Python\\Python39\\site-packages\\transformers\\generation\\configuration_utils.py:601: UserWarning: `do_sample` is set to `False`. However, `temperature` is set to `0.5` -- this flag is only used in sample-based generation modes. You should set `do_sample=True` or unset `temperature`.\n",
      "  warnings.warn(\n",
      "C:\\Users\\robotics\\AppData\\Roaming\\Python\\Python39\\site-packages\\transformers\\generation\\configuration_utils.py:601: UserWarning: `do_sample` is set to `False`. However, `temperature` is set to `1.5` -- this flag is only used in sample-based generation modes. You should set `do_sample=True` or unset `temperature`.\n",
      "  warnings.warn(\n",
      "C:\\Users\\robotics\\AppData\\Roaming\\Python\\Python39\\site-packages\\transformers\\generation\\configuration_utils.py:606: UserWarning: `do_sample` is set to `False`. However, `top_p` is set to `0.9` -- this flag is only used in sample-based generation modes. You should set `do_sample=True` or unset `top_p`.\n",
      "  warnings.warn(\n",
      "C:\\Users\\robotics\\AppData\\Roaming\\Python\\Python39\\site-packages\\transformers\\generation\\configuration_utils.py:623: UserWarning: `do_sample` is set to `False`. However, `top_k` is set to `30` -- this flag is only used in sample-based generation modes. You should set `do_sample=True` or unset `top_k`.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "--- Story with settings: T=0.5, k=50, p=1.0 ---\n",
      "Once upon a time, in a small town nestled between rolling hills and lush greenery, there lived a young girl named Lily. Lily was a curious and adventurous soul, always seeking new experiences and knowledge. She had a deep love for nature and spent most of her days exploring the nearby forests and meadows.\n",
      "\n",
      "One sunny morning, as Lily ventured deeper into the woods, she stumbled upon a hidden clearing. In the center of the clearing stood a magnificent oak tree, its branches stretching towards the sky like outstretched arms. Intrigued, Lily approached the tree and noticed a small, weathered sign hanging from one of its branches. It read, \"The Enchanted Garden: A Haven of Wonders.\"\n",
      "\n",
      "Driven by her insatiable curiosity, Lily followed the sign's instructions and found herself standing at the entrance of a magical garden\n",
      "\n",
      "\n",
      "--- Story with settings: T=1.0, k=50, p=1.0 ---\n",
      "Once upon a time, in a small town nestled between rolling hills and lush greenery, there lived a young girl named Lily. Lily was a curious and adventurous soul, always seeking new experiences and knowledge. She had a deep love for nature and spent most of her days exploring the nearby forests and meadows.\n",
      "\n",
      "One sunny morning, as Lily ventured deeper into the woods, she stumbled upon a hidden clearing. In the center of the clearing stood a magnificent oak tree, its branches stretching towards the sky like outstretched arms. Intrigued, Lily approached the tree and noticed a small, weathered sign hanging from one of its branches. It read, \"The Enchanted Garden: A Haven of Wonders.\"\n",
      "\n",
      "Driven by her insatiable curiosity, Lily followed the sign's instructions and found herself standing at the entrance of a magical garden\n",
      "\n",
      "\n",
      "--- Story with settings: T=1.5, k=50, p=1.0 ---\n",
      "Once upon a time, in a small town nestled between rolling hills and lush greenery, there lived a young girl named Lily. Lily was a curious and adventurous soul, always seeking new experiences and knowledge. She had a deep love for nature and spent most of her days exploring the nearby forests and meadows.\n",
      "\n",
      "One sunny morning, as Lily ventured deeper into the woods, she stumbled upon a hidden clearing. In the center of the clearing stood a magnificent oak tree, its branches stretching towards the sky like outstretched arms. Intrigued, Lily approached the tree and noticed a small, weathered sign hanging from one of its branches. It read, \"The Enchanted Garden: A Haven of Wonders.\"\n",
      "\n",
      "Driven by her insatiable curiosity, Lily followed the sign's instructions and found herself standing at the entrance of a magical garden\n",
      "\n",
      "\n",
      "--- Story with settings: T=1.0, k=30, p=0.9 ---\n",
      "Once upon a time, in a small town nestled between rolling hills and lush greenery, there lived a young girl named Lily. Lily was a curious and adventurous soul, always seeking new experiences and knowledge. She had a deep love for nature and spent most of her days exploring the nearby forests and meadows.\n",
      "\n",
      "One sunny morning, as Lily ventured deeper into the woods, she stumbled upon a hidden clearing. In the center of the clearing stood a magnificent oak tree, its branches stretching towards the sky like outstretched arms. Intrigued, Lily approached the tree and noticed a small, weathered sign hanging from one of its branches. It read, \"The Enchanted Garden: A Haven of Wonders.\"\n",
      "\n",
      "Driven by her insatiable curiosity, Lily followed the sign's instructions and found herself standing at the entrance of a magical garden\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Note: Environment variable`HF_TOKEN` is set and is the current active token independently from the token you've just configured.\n"
     ]
    }
   ],
   "source": [
    "# 1. Setup\n",
    "import os\n",
    "import torch\n",
    "from huggingface_hub import login\n",
    "from transformers import AutoTokenizer, AutoModelForCausalLM\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "# Fetch the token from environment\n",
    "hf_token = os.environ.get('TOKEN')\n",
    "\n",
    "# Login to Hugging Face using the secret token\n",
    "login(token=hf_token)\n",
    "\n",
    "# Load the models and tokenizers\n",
    "models = {\n",
    "    'Phi-3.5-mini-instruct': {\n",
    "        'tokenizer': AutoTokenizer.from_pretrained(\"microsoft/Phi-3.5-mini-instruct\"),\n",
    "        'model': AutoModelForCausalLM.from_pretrained(\"microsoft/Phi-3.5-mini-instruct\")\n",
    "    }\n",
    "}\n",
    "\n",
    "# Assign the tokenizer and model to variables for use\n",
    "tokenizer = models['Phi-3.5-mini-instruct']['tokenizer']\n",
    "model = models['Phi-3.5-mini-instruct']['model']\n",
    "\n",
    "\n",
    "# 3. Generate Story with Different Parameters\n",
    "def generate_story(prompt=\"Once upon a time\", max_length=200, temperature=1.0, top_k=50, top_p=1.0):\n",
    "    inputs = tokenizer(prompt, return_tensors='pt')\n",
    "    outputs = model.generate(\n",
    "        inputs['input_ids'],\n",
    "        max_length=max_length,\n",
    "        temperature=temperature,  # Controls randomness\n",
    "        top_k=top_k,              # Limits the number of highest probability vocabulary tokens\n",
    "        top_p=top_p               # Nucleus sampling: limits tokens to the most probable ones\n",
    "    )\n",
    "    story = tokenizer.decode(outputs[0], skip_special_tokens=True)\n",
    "    return story\n",
    "\n",
    "# Experiment settings\n",
    "parameters = [\n",
    "    {'temperature': 0.5, 'top_k': 50, 'top_p': 1.0},   # More deterministic output\n",
    "    {'temperature': 1.0, 'top_k': 50, 'top_p': 1.0},   # Default settings\n",
    "    {'temperature': 1.5, 'top_k': 50, 'top_p': 1.0},   # More creative but less coherent\n",
    "    {'temperature': 1.0, 'top_k': 30, 'top_p': 0.9},   # Balanced creativity and coherence\n",
    "]\n",
    "\n",
    "# Collect results\n",
    "results = {}\n",
    "for param in parameters:\n",
    "    story = generate_story(prompt=\"Once upon a time\", max_length=200, **param)\n",
    "    results[f\"T={param['temperature']}, k={param['top_k']}, p={param['top_p']}\"] = story\n",
    "\n",
    "# 4. Display Results\n",
    "for key, value in results.items():\n",
    "    print(f\"\\n--- Story with settings: {key} ---\\n{value}\\n\")"
   ]
  },
  {
   "cell_type": "raw",
   "id": "c75ce373",
   "metadata": {},
   "source": [
    "The lower temperature (0.5) promotes more deterministic and coherent outputs, reducing randomness. This setting helps maintain the quality of storytelling while still allowing for some variability.\n",
    "A temperature of 1.0 allows for a balanced level of randomness and creativity, which works well for this kind of narrative. The consistent output indicates that the model has a stable understanding of the prompt and context.\n",
    "A higher temperature (1.5) is expected to introduce more creativity and variability.\n",
    "With a lower top-k value (30) and a top-p of 0.9, the model has a more limited vocabulary to choose from while still allowing for some randomness. \n",
    "\n",
    "Despite the variations in temperature, top-k, and top-p settings, the generated stories remained consistent.\n",
    "Lower temperatures lead to more coherent outputs, while higher temperatures should theoretically increase creativity. However, the results here indicate that even at higher temperatures, the model's understanding of the narrative was strong enough to produce similar outputs. Adjusting top-k to limit vocabulary didn’t lead to different stories. This might imply that the model's understanding of the context was robust enough that the selections available did not significantly affect the narrative direction."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "220e7ef6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "31ce2214d43248bba26282e6406e865b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "VBox(children=(HTML(value='<center> <img\\nsrc=https://huggingface.co/front/assets/huggingface_logo-noborder.sv…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d696486d698b40ada9fb1168e2a31a4b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Loading checkpoint shards:   0%|          | 0/2 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "--- Story with Greedy Strategy ---\n",
      "Once upon a time, in a small town nestled between rolling hills and lush greenery, there lived a young girl named Lily. Lily was a curious and adventurous soul, always seeking new experiences and knowledge. She had a deep love for nature and spent most of her days exploring the nearby forests and meadows.\n",
      "\n",
      "One sunny morning, as Lily ventured deeper into the woods, she stumbled upon a hidden clearing. In the center of the clearing stood a magnificent oak tree, its branches stretching towards the sky like outstretched arms. Intrigued, Lily approached the tree and noticed a small, weathered sign hanging from one of its branches. It read, \"The Enchanted Garden: A Haven of Wonders.\"\n",
      "\n",
      "Driven by her insatiable curiosity, Lily followed the sign's instructions and found herself standing at the entrance of a magical garden\n",
      "\n",
      "\n",
      "--- Story with Beam Strategy ---\n",
      "Once upon a time, in a small town nestled in the heart of the countryside, there lived a young girl named Lily. Lily was a bright and curious child, always eager to learn new things and explore the world around her.\n",
      "\n",
      "One day, as she was walking through the town square, she noticed a group of people gathered around a man who was speaking passionately about the importance of recycling. Intrigued, Lily approached the group and listened intently as the man explained how recycling could help reduce waste, conserve resources, and protect the environment.\n",
      "\n",
      "Lily was fascinated by the idea of recycling and decided to learn more about it. She visited the local library and checked out books on the topic, eager to understand how she could make a difference in her own community.\n",
      "\n",
      "As she delved deeper into the world of recycling, Lily began to realize just how much waste was\n",
      "\n",
      "\n",
      "--- Story with Top_k Strategy ---\n",
      "Once upon a time, in a small town nestled between rolling hills and lush greenery, there lived a young girl named Lily. Lily was a curious and adventurous soul, always seeking new experiences and knowledge. She had a deep love for nature and spent most of her days exploring the nearby forests and meadows.\n",
      "\n",
      "One sunny morning, as Lily ventured deeper into the woods, she stumbled upon a hidden clearing. In the center of the clearing stood a magnificent oak tree, its branches stretching towards the sky like outstretched arms. Intrigued, Lily approached the tree and noticed a small, weathered sign hanging from one of its branches. It read, \"The Enchanted Garden: A Haven of Wonders.\"\n",
      "\n",
      "Driven by her insatiable curiosity, Lily followed the sign's instructions and found herself standing at the entrance of a magical garden\n",
      "\n",
      "\n",
      "--- Story with Top_p Strategy ---\n",
      "Once upon a time, in a small town nestled between rolling hills and lush greenery, there lived a young girl named Lily. Lily was a curious and adventurous soul, always seeking new experiences and knowledge. She had a deep love for nature and spent most of her days exploring the nearby forests and meadows.\n",
      "\n",
      "One sunny morning, as Lily ventured deeper into the woods, she stumbled upon a hidden clearing. In the center of the clearing stood a magnificent oak tree, its branches stretching towards the sky like outstretched arms. Intrigued, Lily approached the tree and noticed a small, weathered sign hanging from one of its branches. It read, \"The Enchanted Garden: A Haven of Wonders.\"\n",
      "\n",
      "Driven by her insatiable curiosity, Lily followed the sign's instructions and found herself standing at the entrance of a magical garden\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Note: Environment variable`HF_TOKEN` is set and is the current active token independently from the token you've just configured.\n"
     ]
    }
   ],
   "source": [
    "# 1. Setup\n",
    "import os\n",
    "import torch\n",
    "from huggingface_hub import login\n",
    "from transformers import AutoTokenizer, AutoModelForCausalLM\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "# Fetch the token from environment\n",
    "hf_token = os.environ.get('TOKEN')\n",
    "\n",
    "# Login to Hugging Face using the secret token\n",
    "login(token=hf_token)\n",
    "\n",
    "# Load the models and tokenizers\n",
    "models = {\n",
    "    'Phi-3.5-mini-instruct': {\n",
    "        'tokenizer': AutoTokenizer.from_pretrained(\"microsoft/Phi-3.5-mini-instruct\"),\n",
    "        'model': AutoModelForCausalLM.from_pretrained(\"microsoft/Phi-3.5-mini-instruct\")\n",
    "    }\n",
    "}\n",
    "\n",
    "# Assign the tokenizer and model to variables for use\n",
    "tokenizer = models['Phi-3.5-mini-instruct']['tokenizer']\n",
    "model = models['Phi-3.5-mini-instruct']['model']\n",
    "\n",
    "\n",
    "# 3. Generate Story with Different Strategies\n",
    "def generate_story(prompt=\"Once upon a time\", max_length=200, strategy=\"greedy\", temperature=1.0, top_k=50, top_p=1.0):\n",
    "    inputs = tokenizer(prompt, return_tensors='pt')\n",
    "    \n",
    "    if strategy == \"greedy\":\n",
    "        outputs = model.generate(inputs['input_ids'], max_length=max_length)\n",
    "    elif strategy == \"beam\":\n",
    "        outputs = model.generate(inputs['input_ids'], max_length=max_length, num_beams=5, early_stopping=True)\n",
    "    elif strategy == \"top_k\":\n",
    "        outputs = model.generate(\n",
    "            inputs['input_ids'],\n",
    "            max_length=max_length,\n",
    "            temperature=temperature,\n",
    "            top_k=top_k\n",
    "        )\n",
    "    elif strategy == \"top_p\":\n",
    "        outputs = model.generate(\n",
    "            inputs['input_ids'],\n",
    "            max_length=max_length,\n",
    "            temperature=temperature,\n",
    "            top_p=top_p\n",
    "        )\n",
    "    \n",
    "    story = tokenizer.decode(outputs[0], skip_special_tokens=True)\n",
    "    return story\n",
    "\n",
    "# Experiment settings\n",
    "strategies = [\n",
    "    {\"strategy\": \"greedy\"},      # Greedy Search\n",
    "    {\"strategy\": \"beam\"},        # Beam Search\n",
    "    {\"strategy\": \"top_k\", \"temperature\": 1.0, \"top_k\": 50},  # Top-k Sampling\n",
    "    {\"strategy\": \"top_p\", \"temperature\": 1.0, \"top_p\": 0.9}   # Top-p Sampling\n",
    "]\n",
    "\n",
    "# Collect results\n",
    "results = {}\n",
    "for params in strategies:\n",
    "    story = generate_story(prompt=\"Once upon a time\", max_length=200, **params)\n",
    "    results[f\"{params['strategy'].capitalize()}\"] = story\n",
    "\n",
    "# 4. Display Results\n",
    "for key, value in results.items():\n",
    "    print(f\"\\n--- Story with {key} Strategy ---\\n{value}\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "8cbdb6ef",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Metrics for Generated Stories:\n",
      "        Type-Token Ratio  Perplexity\n",
      "Greedy          0.773723    2.652047\n",
      "Beam            0.644737    2.657604\n",
      "Top_k           0.773723    2.652047\n",
      "Top_p           0.773723    2.652047\n"
     ]
    }
   ],
   "source": [
    "def compute_metrics(text):\n",
    "    tokens = text.split()\n",
    "    type_token_ratio = len(set(tokens)) / len(tokens) if len(tokens) > 0 else 0\n",
    "    perplexity = np.exp(-np.mean([np.log((len(tokens) - i) / len(tokens)) for i in range(len(tokens))]))\n",
    "\n",
    "    return type_token_ratio, perplexity\n",
    "\n",
    "# Collect metrics for each generated story\n",
    "metrics_results = {}\n",
    "for key, story in results.items():\n",
    "    metrics = compute_metrics(story)\n",
    "    metrics_results[key] = metrics\n",
    "\n",
    "# Display the metrics\n",
    "metrics_df = pd.DataFrame.from_dict(metrics_results, orient='index', columns=['Type-Token Ratio', 'Perplexity'])\n",
    "print(\"\\nMetrics for Generated Stories:\")\n",
    "print(metrics_df)"
   ]
  },
  {
   "cell_type": "raw",
   "id": "dfc9f0eb",
   "metadata": {},
   "source": [
    "Greedy and sampling methods (top-k and top-p) yielded the most diverse and coherent stories, while beam search, although focused, produced slightly less diversity."
   ]
  }
 ],
 "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": "3.9.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}