ibm-granite-test.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import torch\n",
    "from transformers import AutoModelForCausalLM, AutoTokenizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "HUGGINGFACE_API_KEY = os.environ.get('API_KEY')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "device = 'cuda' if torch.cuda.is_available() else 'cpu'\n",
    "model_id = 'ibm-granite/granite-8b-code-base-128k'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "tokenizer = AutoTokenizer.from_pretrained(model_id, cache_dir=\"./models/\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.cuda.empty_cache()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = AutoModelForCausalLM.from_pretrained(model_id, device_map=device, cache_dir=\"./models/\")\n",
    "model.eval()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Change input text\n",
    "input_text = \"def fibonacci():\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Tokenize the text\n",
    "input_tokens = tokenizer(input_text, return_tensors=\"pt\") # Return set for PyTorch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Transfer tokenized inputs to the device\n",
    "for i in input_tokens:\n",
    "    input_tokens[i] = input_tokens[i].to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Generate output tokens with a specified maximum number of new tokens\n",
    "output = model.generate(\n",
    "    **input_tokens,\n",
    "    max_new_tokens=50,  # Limit the number of tokens to generate\n",
    "    do_sample=False,    # Use greedy decoding for faster results\n",
    "    num_beams=1,        # Beam search size; 1 means no beam search\n",
    "    early_stopping=True # Stop early when possible\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Decode tokens into text\n",
    "output = tokenizer.batch_decode(output)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Loop over the batch to print, in this example the batch size is 1\n",
    "for i in output:\n",
    "    print(i)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "tesis",
   "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": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}