Upload latin_abbreviation_expansion.ipynb

latin_abbreviation_expansion.ipynb

@@ -0,0 +1,199 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "1f175efa",
+   "metadata": {},
+   "source": [
+    "# Latin Abbreviation Expansion\n",
+    "\n",
+    "This notebook demonstrates how to use the byt5 model `mschonhardt/abbreviationes-v2`.\n",
+    "It expands medieval abbreviations based on a fixed set of special characters.\n",
+    "\n",
+    "## Quick check\n",
+    "You can use `pipeline` to quickly convert input text. "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 13,
+   "id": "1cd29ad2",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Device set to use cuda:0\n",
+      "Both `max_new_tokens` (=256) and `max_length`(=512) seem to have been set. `max_new_tokens` will take precedence. Please refer to the documentation for more information. (https://huggingface.co/docs/transformers/main/en/main_classes/text_generation)\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Source: aut ferrum lapsū de manubrio\n",
+      "Expanded: aut ferrum lapsum de manubrio\n"
+     ]
+    }
+   ],
+   "source": [
+    "from transformers import pipeline\n",
+    "\n",
+    "# Load the expander\n",
+    "expander = pipeline(\"text2text-generation\", model=\"mschonhardt/abbreviationes-v2\")\n",
+    "\n",
+    "# Example: \"aut ferrum lapsū de manubrio\" abbreviated\n",
+    "text = \"aut ferrum lapsū de manubrio\"\n",
+    "result = expander(text, max_length=512)\n",
+    "\n",
+    "print(f\"Source: {text}\")\n",
+    "print(f\"Expanded: {result[0]['generated_text']}\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "b87f3e45",
+   "metadata": {},
+   "source": [
+    "The model can also be used and exemplified in a more detailed way. \n",
+    "\n",
+    "## Setup Environment"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 14,
+   "id": "044ae4ef",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Torch version: 2.10.0+cu128\n",
+      "Device: cuda\n",
+      "Environment ready.\n"
+     ]
+    }
+   ],
+   "source": [
+    "# Import necessary libraries\n",
+    "import torch\n",
+    "from transformers import AutoTokenizer, AutoModelForSeq2SeqLM\n",
+    "\n",
+    "# Model should be used with GPU (cuda) if available for faster inference\n",
+    "device = \"cuda\" if torch.cuda.is_available() else \"cpu\"\n",
+    "\n",
+    "print(f\"Torch version: {torch.__version__}\")\n",
+    "print(f\"Device: {device}\")\n",
+    "\n",
+    "print(\"Environment ready.\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "4de2def2",
+   "metadata": {},
+   "source": [
+    "## Load the Model from Hugging Face"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 15,
+   "id": "aa5810a8",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Loading model: mschonhardt/abbreviationes-v2 ...\n",
+      "Model loaded successfully!\n"
+     ]
+    }
+   ],
+   "source": [
+    "# Load the model and tokenizer from Huggingface\n",
+    "model_name = \"mschonhardt/abbreviationes-v2\" \n",
+    "print(f\"Loading model: {model_name} ...\")\n",
+    "tokenizer = AutoTokenizer.from_pretrained(model_name, use_fast=False)\n",
+    "model = AutoModelForSeq2SeqLM.from_pretrained(model_name).to(device)\n",
+    "print(\"Model loaded successfully!\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "2dd05d72",
+   "metadata": {},
+   "source": [
+    "### Prediction Logic\n",
+    "The model was trained on abbreviated text lines from manuscripts. Quality might degrade if used for longer passages.\n",
+    "\n",
+    "### Run Inference"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "id": "e858df99",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Input:    aut ferrum lapsū de manubrio\n",
+      "Expanded: aut ferrum lapsum de manubrio\n",
+      "Input:    ei᷒ et surgens ꝑcusserit eum et\n",
+      "Expanded: eius et surgens percusserit eum et\n",
+      "Input:    tur ab ultore sanguinis ꝓximi sui\n",
+      "Expanded: tur ab ultore sanguinis proximi sui\n",
+      "Input:    et illū qui armis c̅tra iniquitatē\n",
+      "Expanded: et illum qui armis contra iniquitatem\n"
+     ]
+    }
+   ],
+   "source": [
+    "# The abbreviated Medieval Latin text\n",
+    "lines = [\"aut ferrum lapsū de manubrio\", \"ei᷒ et surgens ꝑcusserit eum et\", \"tur ab ultore sanguinis ꝓximi sui\", \"et illū qui armis c̅tra iniquitatē\"]\n",
+    "\n",
+    "for input_text in lines:\n",
+    "\n",
+    "    # 1. Tokenize input\n",
+    "    inputs = tokenizer(input_text, return_tensors=\"pt\").to(device)\n",
+    "\n",
+    "    # 2. Generate output tokens\n",
+    "    output_tokens = model.generate(**inputs, max_length=128)\n",
+    "\n",
+    "    # 3. Decode back to text\n",
+    "    expanded_text = tokenizer.decode(output_tokens[0], skip_special_tokens=True)\n",
+    "\n",
+    "    print(f\"Input:    {input_text}\")\n",
+    "    print(f\"Expanded: {expanded_text}\")\n"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "venv-jupyter",
+   "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": "3.12.3"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}