Nandoguer commited on
Commit
17104d0
·
verified ·
1 Parent(s): 11d22af

Upload ReadMe_ForFinalExamNLPAPP.txt

Browse files
Files changed (1) hide show
  1. ReadMe_ForFinalExamNLPAPP.txt +37 -0
ReadMe_ForFinalExamNLPAPP.txt ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ NLP Image Quote Generator APP README
2
+ Overview
3
+ This project aims to classify textual data into emotion categories using various models, including RoBERTa, DistilBERT, and a Bag-of-Words (BoW) classifier.
4
+ Once the emotion has been categorized it uses said emotion to bring a quote from a different dataset than the one used for training and an image from
5
+ stable diffusion.
6
+
7
+ This README provides instructions on setting up and running the code, along with the required libraries.
8
+
9
+ Prerequisites
10
+ Before running the code, ensure you have the following installed:
11
+
12
+ Python 3.6 or later
13
+ pip (Python Package Installer)
14
+ Required Libraries
15
+ The project relies on several Python librariesas follows:
16
+
17
+ from transformers import pipeline, RobertaTokenizer, RobertaForSequenceClassification
18
+ from sklearn.feature_extraction.text import CountVectorizer
19
+ from sklearn.naive_bayes import MultinomialNB
20
+ from sklearn.pipeline import make_pipeline
21
+ from datasets import load_dataset
22
+ from PIL import Image, ImageDraw, ImageFont
23
+ import textwrap
24
+ import random
25
+ from diffusers import StableDiffusionPipeline
26
+ import torch
27
+ from sklearn.metrics import classification_report, accuracy_score, f1_score
28
+ from sklearn.model_selection import train_test_split # Import train_test_split
29
+ import gradio as gr
30
+
31
+ Additionally, we must have access to two separate datasets from the huggingface.-
32
+ https://huggingface.co/datasets/dair-ai/emotion
33
+ https://huggingface.co/datasets/Abirate/english_quotes
34
+
35
+ Evaluation Metrics
36
+ The code evaluates the models based on accuracy, precision, recall, and F1 score. The results are printed out on the script but not the gradio site (to keep
37
+ it clean).