Update app.py

app.py

@@ -1,46 +1,37 @@
-from transformers import pipeline
-from transformers import AutoTokenizer, AutoModelForSequenceClassification
-
-import torch
-import torch.nn.functional as F
-
-sentimentAnalysisModelName = "distilbert-base-uncased-finetuned-sst-2-english"
-sentimentAnalysisModel = AutoModelForSequenceClassification.from_pretrained(sentimentAnalysisModelName)
+import streamlit as st
+from transformers import pipeline, AutoTokenizer
+
+# Define a list of pretrained models
+sentimentAnalysisModels = {
+    "Roberta": "deepset/roberta-base-squad2",
+    "RESPIN": "RESPIN/Telugu_LanguageModels_for_ASR",
+    "GPT2-n": "DunnBC22/gpt2-Causal_Language_Model-AG_News",
+    "Roberta-Base (ADDITIONAL MODEL!)": "achimoraites/TextClassification-roberta-base_ag_news",
+}
+
+# Display a selection box for the user to choose a model
+sentimentAnalysisLanguageSelections = st.selectbox("Please select one of these finetuned models from the dropdown", list(sentimentAnalysisModels.keys()))
+
+# roBERTa specific label map
+languageMapRoberta = {"LABEL_0": "NEGATIVE",
+                     "LABEL_1": "POSITIVE", "LABEL_2": "NEUTRAL"}
+
+# Load the selected model and tokenizer
+sentimentAnalysisModelName = sentimentAnalysisModels[sentimentAnalysisSelectedModels]
 sentimentAnalysisTokenizer = AutoTokenizer.from_pretrained(sentimentAnalysisModelName)
-
-sentimentAnalysisClassifier = pipeline("sentiment-analysis", model = sentimentAnalysisModelName, tokenizer = sentimentAnalysisTokenizer)
-sentimentAnalysisResults = sentimentAnalysisClassifier(["I am very proud to be in NJIT today and studying Data Science.", "I hope you don't hate this food!"])
-
-for sentimentAnalysisResult in sentimentAnalysisResults:
-  print(sentimentAnalysisResult)
-
-sentimentAnalysisTokens = sentimentAnalysisTokenizer.tokenize("I am very proud to be in NJIT today and studying Data Science.")
-sentimentAnalysisTokenids = sentimentAnalysisTokenizer.convert_tokens_to_ids(sentimentAnalysisTokens)
-sentimentAnalysisInputids = sentimentAnalysisTokenizer("I am very proud to be in NJIT today and studying Data Science.")
-
-print(f' Tokens: {sentimentAnalysisTokens}')
-print(f' Token IDs: {sentimentAnalysisTokenids}')
-print(f' Input IDs: {sentimentAnalysisInputids}')
-
-sentimentAnalysisXTrainingVariable = ["I am very proud to be in NJIT today and studying Data Science.", "I hope you don't hate this food!"]
-sentimentAnalysisBatch = sentimentAnalysisTokenizer(sentimentAnalysisXTrainingVariable, padding = True, truncation = True, max_length = 512, return_tensors = "pt")
-print(sentimentAnalysisBatch)
-
-with torch.no_grad():
-  sentimentAnalysisOutputs = sentimentAnalysisModel(**sentimentAnalysisBatch, labels = torch.tensor([1, 0]))
-  print(sentimentAnalysisOutputs)
-  sentimentAnalysisPredictions = F.softmax(sentimentAnalysisOutputs.logits, dim = 1)
-  print(sentimentAnalysisPredictions)
-  sentimentAnalysisLabels = torch.argmax(sentimentAnalysisPredictions, dim = 1)
-  print(sentimentAnalysisLabels)
-  sentimentAnalysisLabels = [sentimentAnalysisModel.config.id2label[sentimentAnalysisLabelID] for sentimentAnalysisLabelID in sentimentAnalysisLabels.tolist()]
-  print(sentimentAnalysisLabels)
-
-sentimentAnalysis_saveDirectory = "modelSaved"
-sentimentAnalysisTokenizer.save_pretrained(sentimentAnalysis_saveDirectory)
-sentimentAnalysisModel.save_pretrained(sentimentAnalysis_saveDirectory)
-
-sentimentAnalysisTokenizer = AutoTokenizer.from_pretrained(sentimentAnalysis_saveDirectory)
-sentimentAnalysisModel = AutoModelForSequenceClassification.from_pretrained(sentimentAnalysis_saveDirectory)
-
-
+sentiment_pipeline = pipeline(
+    "sentiment-analysis", model=sentimentAnalysisModelName, tokenizer=sentimentAnalysisTokenizer)
+
+# Get user input and perform sentiment analysis
+sentimentAnalysisTextInput = st.text_input("Please enter sample text for finetuned language model below:",
+                           "I am grateful for Data Science Program at NJIT and to have amazing faculty to learn under!")
+sentimentAnalysisSubmitButton = st.button("Press the submit button for final grading")
+
+if sentimentAnalysisSubmitButton and sentimentAnalysisTextInput:
+    sentimentAnalysisFinalOutput = sentiment_pipeline(sentimentAnalysisTextInput)
+    if sentimentAnalysisSelectedModels == "roBERTa":
+        st.write("Roberta Sentiment Analysis Resultant Value:", roberta_label_map[result[0]["Roberta Label"]])
+        st.write("Roberta Sentiment Analysis Resultant Score:", sentimentAnalysisFinalOutput[0]["Probability Assigned with Sentiment Analysis Label"])
+    else:
+        st.write("Sentiment Analysis Resultant Value:", sentimentAnalysisFinalOutput[0]["Sentiment Analysis Label"])
+        st.write("Sentiment Analysis Resultant Score:", sentimentAnalysisFinalOutput[0]["Probability Assigned with Sentiment Analysis Label"])