{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "
\n", "\n", "## [S. Mussard](https://sites.google.com/view/cv-stphane-mussard/accueil \"Homepage\")\n", "\n", "# UM6P\n", "\n", "# Natural Language Processing: LOGIT\n", "\n", "\n", "
\n", "\n", "\n", "\n", "
\n", "\"Scikit-Learn\"/\n", "
\n", "
\"Python\"/ \n", "
\n", " \n" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "
\n", "\n", "## Sentiment Analysis" ] }, { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "# Importation \n", "\n", "%matplotlib inline \n", "import numpy as np\n", "import pandas as pd\n", "import matplotlib.pyplot as plt\n", "from sklearn import metrics\n", "import torch\n", "from torch.utils.data import Dataset, DataLoader\n", "from transformers import AutoModel, AutoTokenizer\n", "from transformers import AutoModelForSequenceClassification, AutoTokenizer\n", "\n", "import gradio as gr\n", "from gradio.components import Label" ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "Some weights of the model checkpoint at ./poids were not used when initializing RobertaModel: ['classifier.out_proj.weight', 'classifier.dense.bias', 'classifier.out_proj.bias', 'classifier.dense.weight']\n", "- This IS expected if you are initializing RobertaModel from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).\n", "- This IS NOT expected if you are initializing RobertaModel from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).\n", "Some weights of RobertaModel were not initialized from the model checkpoint at ./poids and are newly initialized: ['roberta.pooler.dense.weight', 'roberta.pooler.dense.bias']\n", "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n" ] } ], "source": [ "path = \"./weights\"\n", "model = AutoModel.from_pretrained(path, trust_remote_code=True)\n", "class CamembertClass(torch.nn.Module):\n", " def __init__(self):\n", " super(CamembertClass, self).__init__()\n", " self.l1 = model\n", " self.dropout = torch.nn.Dropout(0.1)\n", " self.pre_classifier = torch.nn.Linear(1024, 1024)\n", " self.classifier = torch.nn.Linear(1024, 3)\n", "\n", " def forward(self, input_ids, attention_mask, token_type_ids):\n", " output_1 = self.l1(input_ids=input_ids, attention_mask=attention_mask, token_type_ids=token_type_ids)\n", " hidden_state = output_1[0]\n", " pooler = hidden_state[:, 0]\n", " pooler = self.pre_classifier(pooler)\n", " pooler = torch.nn.ReLU()(pooler)\n", " pooler = self.dropout(pooler)\n", " output = self.classifier(pooler)\n", " return output" ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [], "source": [ "#model_gradio = CamembertClass()\n", "path = \"./pytorch_model.bin\"\n", "model = torch.load(path, map_location=\"cpu\")\n", "path_tokenizer = \"./\"\n", "tokenizer = AutoTokenizer.from_pretrained(path_tokenizer)\n" ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [], "source": [ "#pip install pydantic==1.10.7" ] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Running on local URL: http://127.0.0.1:7860\n", "Running on public URL: https://93ecddda8853b625c0.gradio.live\n", "\n", "This share link expires in 72 hours. For free permanent hosting and GPU upgrades, run `gradio deploy` from Terminal to deploy to Spaces (https://huggingface.co/spaces)\n" ] }, { "data": { "text/html": [ "
" ], "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/plain": [] }, "execution_count": 7, "metadata": {}, "output_type": "execute_result" } ], "source": [ "model.eval() # Mettez votre modèle en mode évaluation\n", "\n", "# Fonction d'inférence pour Gradio\n", "def predict(text):\n", " inputs = tokenizer(text, return_tensors=\"pt\", padding=True, truncation=True, max_length=512)\n", " \n", " # Extract necessary inputs for the model\n", " input_ids = inputs['input_ids']\n", " attention_mask = inputs['attention_mask']\n", " token_type_ids = inputs.get('token_type_ids', None) # Some models do not use segment IDs\n", " \n", " # Make prediction\n", " with torch.no_grad():\n", " # Directly use outputs if your model returns logits directly\n", " logits = model(input_ids=input_ids, attention_mask=attention_mask, token_type_ids=token_type_ids)\n", "\n", " \n", " # Convert logits to probabilities\n", " probabilities = torch.softmax(logits, dim=1).detach().cpu().numpy()[0]\n", " # Replace the following with your actual classes\n", " classes = ['Negative Sentiment', 'Positive Sentiment']\n", " return {classes[i]: float(probabilities[i]) for i in range(len(classes))}\n", "\n", "# Création de l'interface Gradio\n", "iface = gr.Interface(fn=predict,\n", " inputs=gr.components.Textbox(placeholder=\"Enter your text here...\"),\n", " outputs=gr.components.Label(num_top_classes=2))\n", "iface.launch(share=True)\n" ] } ], "metadata": { "hide_input": false, "kernelspec": { "display_name": "Python 3", "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": "3.7.8" }, "toc": { "base_numbering": 1, "nav_menu": { "height": "244px", "width": "252px" }, "number_sections": true, "sideBar": true, "skip_h1_title": false, "title_cell": "Table of Contents", "title_sidebar": "Contents", "toc_cell": false, "toc_position": {}, "toc_section_display": "block", "toc_window_display": false } }, "nbformat": 4, "nbformat_minor": 1 }