Merge pull request #4 from cgr28/milestone-3

.gitignore

@@ -127,3 +127,4 @@ dmypy.json
 
 # Pyre type checker
 .pyre/
+data/

app.py

@@ -1,29 +1,31 @@
 import streamlit as st
-from transformers import AutoTokenizer, RobertaForSequenceClassification
+from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline
 import numpy as np
 import torch
+import pandas as pd
+import torch.nn.functional as F
 
-st.title("CS482 Project Sentiment Analysis")
+model_name = "unitary/toxic-bert"
+
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForSequenceClassification.from_pretrained(model_name)
+
+
+df = pd.DataFrame(columns=("Tweet", "Toxicity", "Probability"))
 
-text = st.text_area(label="Text to be analyzed", value="This sentiment analysis app is great!")
+sample_tweets = ["Ask Sityush to clean up his behavior than issue me nonsensical warnings...", "be a man and lets discuss it-maybe over the phone?", "Don't look, come or think of comming back! Tosser."]
 
-selected_model = st.radio(label="Model", options=["Model 1", "Model 2"])
+classifier = pipeline("sentiment-analysis", model=model, tokenizer=tokenizer)
+results = classifier(sample_tweets)
+
+batch  = tokenizer(sample_tweets, padding=True, truncation=True, max_length=512, return_tensors="pt")
+
+# assignment 3
+st.title("CS482 Project Sentiment Analysis")
 
-analyze_button = st.button(label="Analyze")
+st.markdown("**:red[unitary/toxic-bert]**")
 
-st.markdown("**:red[Sentiment:]**")
+for i in range(len(sample_tweets)):
+    df.loc[len(df.index)] = [sample_tweets[i], results[i]["label"], results[i]["score"]]
 
-with st.spinner(text="Analyzing..."):
-    if analyze_button:
-        if selected_model=="Model 1":
-            tokenizer = AutoTokenizer.from_pretrained("cardiffnlp/twitter-roberta-base-emotion")
-            model = RobertaForSequenceClassification.from_pretrained("cardiffnlp/twitter-roberta-base-emotion")
-        else:
-            tokenizer = AutoTokenizer.from_pretrained("cardiffnlp/twitter-roberta-base-sentiment-latest")
-            model = RobertaForSequenceClassification.from_pretrained("cardiffnlp/twitter-roberta-base-sentiment-latest")
-        inputs = tokenizer(text, return_tensors="pt")
-        with torch.no_grad():
-            logits = model(**inputs).logits
-        prediction_id = logits.argmax().item()
-        results = model.config.id2label[prediction_id]
-        st.write(results)
+st.table(df)

requirements.txt

@@ -5,23 +5,43 @@ cachetools==5.3.0
 certifi==2022.12.7
 charset-normalizer==3.1.0
 click==8.1.3
+cmake==3.26.3
+contourpy==1.0.7
+cycler==0.11.0
 decorator==5.1.1
 emoji==0.6.0
 entrypoints==0.4
 filelock==3.10.6
+fonttools==4.39.3
 gitdb==4.0.10
 GitPython==3.1.31
 huggingface-hub==0.13.3
 idna==3.4
 importlib-metadata==6.1.0
+importlib-resources==5.12.0
 Jinja2==3.1.2
+joblib==1.2.0
 jsonschema==4.17.3
+kiwisolver==1.4.4
+lit==16.0.1
 markdown-it-py==2.2.0
 MarkupSafe==2.1.2
+matplotlib==3.7.1
 mdurl==0.1.2
 mpmath==1.3.0
 networkx==3.0
 numpy==1.24.2
+nvidia-cublas-cu11==11.10.3.66
+nvidia-cuda-cupti-cu11==11.7.101
+nvidia-cuda-nvrtc-cu11==11.7.99
+nvidia-cuda-runtime-cu11==11.7.99
+nvidia-cudnn-cu11==8.5.0.96
+nvidia-cufft-cu11==10.9.0.58
+nvidia-curand-cu11==10.2.10.91
+nvidia-cusolver-cu11==11.4.0.1
+nvidia-cusparse-cu11==11.7.4.91
+nvidia-nccl-cu11==2.14.3
+nvidia-nvtx-cu11==11.7.91
 packaging==23.0
 pandas==1.5.3
 Pillow==9.4.0
@@ -30,6 +50,7 @@ pyarrow==11.0.0
 pydeck==0.8.0
 Pygments==2.14.0
 Pympler==1.0.1
+pyparsing==3.0.9
 pyrsistent==0.19.3
 python-dateutil==2.8.2
 pytz==2023.3
@@ -38,11 +59,15 @@ PyYAML==6.0
 regex==2023.3.23
 requests==2.28.2
 rich==13.3.3
+scikit-learn==1.2.2
+scipy==1.10.1
 semver==3.0.0
 six==1.16.0
+sklearn==0.0.post4
 smmap==5.0.0
 streamlit==1.20.0
 sympy==1.11.1
+threadpoolctl==3.1.0
 tokenizers==0.13.2
 toml==0.10.2
 toolz==0.12.0
@@ -51,9 +76,11 @@ torchvision==0.15.1
 tornado==6.2
 tqdm==4.65.0
 transformers==4.27.4
-typing-extensions==4.5.0
+triton==2.0.0
+typing_extensions==4.5.0
 tzdata==2023.3
 tzlocal==4.3
 urllib3==1.26.15
 validators==0.20.0
+watchdog==3.0.0
 zipp==3.15.0

train.py

@@ -0,0 +1,106 @@
+from transformers import BertTokenizerFast, BertModel, Trainer, TrainingArguments
+import torch
+from torch.utils.data import Dataset
+# from torch.optim import AdamW
+import pandas as pd
+from sklearn.model_selection import train_test_split
+
+
+# assignment 3
+model_name = "bert-base-uncased"
+
+class ToxicDataset(Dataset):
+
+    def __init__(self, encodings, labels):
+        self.encodings = encodings
+        self.labels = labels
+    
+    def __getitem__(self, idx):
+        item = {key: torch.tensor(val[idx]) for key, val in self.encodings.items()}
+        item["labels"] = torch.tensor(self.labels[idx])
+        return item
+
+    def __len__(self):
+        return len(self.labels)
+
+print("Reading data...")
+data = pd.read_csv("./data/train.csv")
+toxic_data = pd.DataFrame()
+toxic_data["text"] = data["comment_text"]
+toxic_data["labels"] = data.iloc[:, 2:].values.tolist()
+
+print("Data read. Splitting data...")
+train_texts, val_texts, train_labels, val_labels = train_test_split(toxic_data.text.to_list(), toxic_data.labels.to_list(), test_size=.2)
+
+
+print("Data split. Tokenizing data...")
+tokenizer = BertTokenizerFast.from_pretrained(model_name)
+
+train_encodings = tokenizer.batch_encode_plus(train_texts, truncation=True, padding=True, return_tensors='pt')
+val_encodings = tokenizer.batch_encode_plus(val_texts, truncation=True, padding=True, return_tensors='pt')
+
+
+train_dataset = ToxicDataset(train_encodings, train_labels)
+val_dataset = ToxicDataset(val_encodings, val_labels)
+
+print("Data tokenized. Beginning training...")
+
+training_args = TrainingArguments(
+    output_dir="./results",          
+    num_train_epochs=2,             
+    per_device_train_batch_size=4, 
+    per_device_eval_batch_size=16, 
+    warmup_steps=500,               
+    weight_decay=0.01,             
+    logging_dir="./logs",            
+    logging_steps=10,
+)
+
+# device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
+
+model = BertModel.from_pretrained(model_name, num_labels=6)
+
+trainer = Trainer(
+    model=model,
+    args=training_args,
+    train_dataset=train_dataset,
+    eval_dataset=val_dataset,
+)
+
+trainer.train()
+
+# model = DistilBertForSequenceClassification.from_pretrained(model_name, num_labels=6)
+
+# model.to(device)
+# model.train()
+
+# train_loader = DataLoader(train_dataset, batch_size=16, shuffle=True)
+
+# optim = AdamW(model.parameters(), lr=5e-5)
+
+# num_train_epochs = 2
+
+# for epoch in range(num_train_epochs):
+#     for batch in train_loader:
+#         optim.zero_grad()
+#         input_ids = batch["input_ids"].to(device)
+#         attention_mask = batch["attention_mask"].to(device)
+#         labels = batch["labels"].to(device)
+
+#         outputs = model(input_ids, attention_mask=attention_mask, labels=labels)
+        
+#         loss = outputs[0]
+#         loss.backward()
+#         optim.step()
+
+# model.eval()
+
+
+
+
+print("Training complete. Saving model...")
+
+save_directory = "./results/model"
+model.save_pretrained(save_directory)
+
+print("Model saved.")