Initial commit

README.md

@@ -0,0 +1,3 @@
+# DIPLOMOVA_PRACA
+
+Diplomova praca Omasta David 

api.py

@@ -0,0 +1,48 @@
+import torch
+import uvicorn
+from fastapi import FastAPI
+from pydantic import BaseModel
+from transformers import  MT5Tokenizer,AutoTokenizer, AutoModel ,T5ForConditionalGeneration
+import warnings
+import json
+import random
+import torch.nn.functional as F
+#from ece import compute_ECE
+from torch.utils.data import DataLoader
+from functools import reduce
+
+warnings.filterwarnings("ignore")
+DEVICE ='cpu'
+model_dir = "C:/Users/david/Desktop/T5_JUPYTER/qa_model"
+tokenizer_dir = "C:/Users/david/Desktop/T5_JUPYTER/qa_tokenizer"
+MODEL = T5ForConditionalGeneration.from_pretrained(model_dir, from_tf=False, return_dict=True).to(DEVICE)
+print("Model succesfully loaded!")
+TOKENIZER = AutoTokenizer.from_pretrained(tokenizer_dir, use_fast=True)
+print("Tokenizer succesfully loaded!")
+Q_LEN = 512
+TOKENIZER.add_tokens('<sep>')
+
+
+print('model loaded')
+
+app = FastAPI()
+
+# BASE MODEL
+class InputData(BaseModel):
+    context: str
+    question: str
+
+@app.post("/predict")
+async def predict(input_data: InputData):
+    inputs = TOKENIZER(input_data.question, input_data.context, max_length=512, padding="max_length", truncation=True, add_special_tokens=True)
+    input_ids = torch.tensor(inputs["input_ids"], dtype=torch.long).to(DEVICE).unsqueeze(0)
+    attention_mask = torch.tensor(inputs["attention_mask"], dtype=torch.long).to(DEVICE).unsqueeze(0)
+    outputs = MODEL.generate(input_ids=input_ids, attention_mask=attention_mask, return_dict_in_generate=True,output_scores=True,max_length=512)
+    predicted_ids = outputs.sequences.numpy()
+    predicted_text = TOKENIZER.decode(predicted_ids[0], skip_special_tokens=True)
+
+    return {'prediction':predicted_text}
+
+
+if __name__ == "__main__":
+    uvicorn.run(app, host="127.0.0.1", port=8090)

aplication.py

@@ -0,0 +1,33 @@
+import requests
+import json
+import streamlit as st
+
+def predict(context,question):
+    url = 'http://localhost:8090/predict'
+    data = {'context': context,'question': question}
+    json_data = json.dumps(data)
+    headers = {'Content-type': 'application/json'}
+    response = requests.post(url, data=json_data, headers=headers)
+    result = response.json()
+    return result
+
+def main():
+    st.title("T5 model inference")
+    
+    # Vytvoríme polia pre zadanie hodnôt
+    context = st.text_input("context:")
+    question = st.text_input("question:")
+    prediction = predict(context,question)
+    # Vytvoríme tlačidlo pre vykonanie akcie
+    if st.button("Execute"):
+        
+        st.json({
+    'context': context,
+    'question': question,
+    'prediciton':prediction
+    })
+        
+
+
+if __name__ == "__main__":
+    main()

condadiplo.yaml

@@ -0,0 +1,91 @@
+name: DIPLOcondaEnviroment
+channels:
+  - defaults
+dependencies:
+  - bzip2=1.0.8=he774522_0
+  - ca-certificates=2023.01.10=haa95532_0
+  - libffi=3.4.2=hd77b12b_6
+  - openssl=1.1.1t=h2bbff1b_0
+  - pip=23.0.1=py311haa95532_0
+  - python=3.11.2=h966fe2a_0
+  - setuptools=66.0.0=py311haa95532_0
+  - sqlite=3.41.2=h2bbff1b_0
+  - tk=8.6.12=h2bbff1b_0
+  - vc=14.2=h21ff451_1
+  - vs2015_runtime=14.27.29016=h5e58377_2
+  - wheel=0.38.4=py311haa95532_0
+  - xz=5.2.10=h8cc25b3_1
+  - zlib=1.2.13=h8cc25b3_0
+  - pip:
+      - altair==4.2.2
+      - anyio==3.6.2
+      - attrs==23.1.0
+      - blinker==1.6.2
+      - cachetools==5.3.0
+      - certifi==2022.12.7
+      - charset-normalizer==3.1.0
+      - click==8.1.3
+      - colorama==0.4.6
+      - comtypes==1.1.14
+      - decorator==5.1.1
+      - entrypoints==0.4
+      - fastapi==0.92.0
+      - filelock==3.11.0
+      - gitdb==4.0.10
+      - gitpython==3.1.31
+      - h11==0.14.0
+      - huggingface-hub==0.12.1
+      - idna==3.4
+      - importlib-metadata==6.4.1
+      - jinja2==3.1.2
+      - jsonschema==4.17.3
+      - markdown-it-py==2.2.0
+      - markupsafe==2.1.2
+      - mdurl==0.1.2
+      - mouseinfo==0.1.3
+      - mpmath==1.3.0
+      - networkx==3.1
+      - numpy==1.24.2
+      - packaging==23.1
+      - pandas==1.5.3
+      - pillow==9.5.0
+      - protobuf==3.20.3
+      - pyarrow==11.0.0
+      - pydantic==1.10.7
+      - pydeck==0.8.1b0
+      - pygments==2.15.0
+      - pympler==1.0.1
+      - pyperclip==1.8.2
+      - pyrsistent==0.19.3
+      - python-dateutil==2.8.2
+      - pytz==2023.3
+      - pytz-deprecation-shim==0.1.0.post0
+      - pywin32==305
+      - pywinauto==0.6.8
+      - pyyaml==6.0
+      - regex==2023.3.23
+      - requests==2.28.2
+      - rich==13.3.4
+      - six==1.16.0
+      - smmap==5.0.0
+      - sniffio==1.3.0
+      - starlette==0.25.0
+      - streamlit==1.21.0
+      - sympy==1.11.1
+      - tokenizers==0.13.2
+      - toml==0.10.2
+      - toolz==0.12.0
+      - torch==2.0.0
+      - torchaudio==2.0.1
+      - torchvision==0.15.1
+      - tornado==6.2
+      - tqdm==4.65.0
+      - transformers==4.26.1
+      - typing-extensions==4.5.0
+      - tzdata==2023.3
+      - tzlocal==4.3
+      - urllib3==1.26.15
+      - uvicorn==0.20.0
+      - validators==0.20.0
+      - watchdog==3.0.0
+      - zipp==3.15.0

requirements.txt

@@ -0,0 +1,72 @@
+altair==4.2.2
+anyio==3.6.2
+attrs==23.1.0
+blinker==1.6.2
+cachetools==5.3.0
+certifi==2022.12.7
+charset-normalizer==3.1.0
+click==8.1.3
+colorama==0.4.6
+comtypes==1.1.14
+decorator==5.1.1
+entrypoints==0.4
+fastapi==0.92.0
+filelock==3.11.0
+gitdb==4.0.10
+GitPython==3.1.31
+h11==0.14.0
+huggingface-hub==0.12.1
+idna==3.4
+importlib-metadata==6.4.1
+Jinja2==3.1.2
+jsonschema==4.17.3
+markdown-it-py==2.2.0
+MarkupSafe==2.1.2
+mdurl==0.1.2
+MouseInfo==0.1.3
+mpmath==1.3.0
+networkx==3.1
+numpy==1.24.2
+packaging==23.1
+pandas==1.5.3
+Pillow==9.5.0
+protobuf==3.20.3
+pyarrow==11.0.0
+pydantic==1.10.7
+pydeck==0.8.1b0
+Pygments==2.15.0
+Pympler==1.0.1
+pyperclip==1.8.2
+pyrsistent==0.19.3
+python-dateutil==2.8.2
+pytz==2023.3
+pytz-deprecation-shim==0.1.0.post0
+pywin32==305
+pywinauto==0.6.8
+PyYAML==6.0
+regex==2023.3.23
+requests==2.28.2
+rich==13.3.4
+six==1.16.0
+smmap==5.0.0
+sniffio==1.3.0
+starlette==0.25.0
+streamlit==1.21.0
+sympy==1.11.1
+tokenizers==0.13.2
+toml==0.10.2
+toolz==0.12.0
+torch==2.0.0
+torchaudio==2.0.1
+torchvision==0.15.1
+tornado==6.2
+tqdm==4.65.0
+transformers==4.26.1
+typing_extensions==4.5.0
+tzdata==2023.3
+tzlocal==4.3
+urllib3==1.26.15
+uvicorn==0.20.0
+validators==0.20.0
+watchdog==3.0.0
+zipp==3.15.0

train.py

@@ -0,0 +1,175 @@
+import torch
+import json
+from tqdm import tqdm
+import torch.nn as nn
+from torch.optim import Adam
+import nltk
+import spacy
+import string
+import evaluate  # Bleu
+from torch.utils.data import Dataset, DataLoader, RandomSampler
+import pandas as pd
+import numpy as np
+import transformers
+from sklearn.model_selection import train_test_split
+import matplotlib.pyplot as plt
+#from transformers import T5Tokenizer, T5Model, T5ForConditionalGeneration, T5TokenizerFast
+from transformers import AutoTokenizer, T5ForConditionalGeneration
+import warnings
+warnings.filterwarnings("ignore")
+
+print("Imports succesfully done")
+
+DEVICE ='cuda:0'
+#TOKENIZER = AutoTokenizer.from_pretrained("ApoTro/slovak-t5-small")
+TOKENIZER=AutoTokenizer.from_pretrained('google/mt5-small')
+#TOKENIZER=AutoTokenizer.from_pretrained('google/mt5-base')
+TOKENIZER.add_tokens('<sep>')
+#MODEL = T5ForConditionalGeneration.from_pretrained("ApoTro/slovak-t5-small").to(DEVICE)
+MODEL = T5ForConditionalGeneration.from_pretrained("google/mt5-small").to(DEVICE)
+#MODEL = T5ForConditionalGeneration.from_pretrained("google/mt5-base").to(DEVICE)
+
+#pridam token 
+MODEL.resize_token_embeddings(len(TOKENIZER))
+#lr = learning rate = 10-5
+OPTIMIZER = Adam(MODEL.parameters(), lr=0.00001)
+Q_LEN = 256   # Question Length
+T_LEN = 32    # Target Length
+BATCH_SIZE = 4  #dávka dát
+print("Model succesfully loaded")
+
+path_train = '/home/omasta/T5_JUPYTER/skquad-221017/train-v1.json' 
+
+with open(path_train) as f:
+    data = json.load(f)
+
+
+def prepare_data(data):
+    articles = []
+    for article in data["data"]:
+        for paragraph in article["paragraphs"]:
+            for qa in paragraph["qas"]:
+                question = qa["question"]
+
+                answer = qa["answers"][0]["text"]
+                #input_ = 'generuj_odpoved : ' + paragraph["context"] + ' <sep>' + question + ' <sep>'
+
+                inputs = {"input": paragraph["context"]+'<sep>'+question, "answer": answer}
+                #inputs = {'context': input_ ,'answer':answer}
+
+                articles.append(inputs)
+
+    return articles
+
+prepared_data = prepare_data(data)
+print(prepared_data[0])
+
+#Dataframe
+data = pd.DataFrame(prepared_data)
+
+
+
+class QA_Dataset(Dataset):
+    def __init__(self, tokenizer, dataframe, q_len, t_len):
+        self.tokenizer = tokenizer
+        self.q_len = q_len
+        self.t_len = t_len
+        self.data = dataframe
+        self.input = self.data['input']
+        #self.context = self.data["context"]
+        self.answer = self.data['answer']
+    def __len__(self):
+        return len(self.questions)
+
+    def __getitem__(self, idx):
+        input = self.input[idx]
+        answer = self.answer[idx]
+
+        input_tokenized = self.tokenizer(input, max_length=self.q_len, padding="max_length",
+                                                    truncation=True, pad_to_max_length=True, add_special_tokens=True)
+        answer_tokenized = self.tokenizer(answer, max_length=self.t_len, padding="max_length", 
+                                          truncation=True, pad_to_max_length=True, add_special_tokens=True)
+
+        labels = torch.tensor(answer_tokenized["input_ids"], dtype=torch.long)
+        labels[labels == 0] = -100
+
+        return {
+            "input_ids": torch.tensor(input_tokenized["input_ids"], dtype=torch.long),
+            "attention_mask": torch.tensor(input_tokenized["attention_mask"], dtype=torch.long),
+            "labels": labels,
+            "decoder_attention_mask": torch.tensor(answer_tokenized["attention_mask"], dtype=torch.long)
+        }
+
+
+
+##DATA LOADERS
+
+train_data, val_data = train_test_split(data, test_size=0.2, random_state=42)
+train_sampler = RandomSampler(train_data.index)
+val_sampler = RandomSampler(val_data.index)
+qa_dataset = QA_Dataset(TOKENIZER, data, Q_LEN, T_LEN)
+
+train_loader = DataLoader(qa_dataset, batch_size=BATCH_SIZE, sampler=train_sampler)
+val_loader = DataLoader(qa_dataset, batch_size=BATCH_SIZE, sampler=val_sampler)
+print("Loaders working fine")
+
+
+
+
+### TRAINING  (46MINS ACCORDING THE V1_DATA)
+
+
+train_loss = 0
+val_loss = 0
+train_batch_count = 0
+val_batch_count = 0
+
+for epoch in range(4):
+    MODEL.train()
+    for batch in tqdm(train_loader, desc="Training batches"):
+        input_ids = batch["input_ids"].to(DEVICE)
+        attention_mask = batch["attention_mask"].to(DEVICE)
+        labels = batch["labels"].to(DEVICE)
+        decoder_attention_mask = batch["decoder_attention_mask"].to(DEVICE)
+
+        outputs = MODEL(
+                          input_ids=input_ids,
+                          attention_mask=attention_mask,
+                          labels=labels,
+                          decoder_attention_mask=decoder_attention_mask
+                        )
+
+        OPTIMIZER.zero_grad()
+        outputs.loss.backward()
+        OPTIMIZER.step()
+        train_loss += outputs.loss.item()
+        train_batch_count += 1
+
+    #Evaluation
+    MODEL.eval()
+    for batch in tqdm(val_loader, desc="Validation batches"):
+        input_ids = batch["input_ids"].to(DEVICE)
+        attention_mask = batch["attention_mask"].to(DEVICE)
+        labels = batch["labels"].to(DEVICE)
+        decoder_attention_mask = batch["decoder_attention_mask"].to(DEVICE)
+
+        outputs = MODEL(
+                          input_ids=input_ids,
+                          attention_mask=attention_mask,
+                          labels=labels,
+                          decoder_attention_mask=decoder_attention_mask
+                        )
+
+        OPTIMIZER.zero_grad()
+        outputs.loss.backward()
+        OPTIMIZER.step()
+        val_loss += outputs.loss.item()
+        val_batch_count += 1 
+    print(f"{epoch+1}/{2} -> Train loss: {train_loss / train_batch_count}\tValidation loss: {val_loss/val_batch_count}")
+
+
+print("Training done succesfully")
+
+## SAVE FINE_TUNED MODEL
+MODEL.save_pretrained("qa_model_mT5_small")
+TOKENIZER.save_pretrained('qa_tokenizer_mT5_small')

usecase.py

@@ -0,0 +1,139 @@
+## IMPORT NESSESARY EQUIPMENTS
+from transformers import T5ForConditionalGeneration, T5Tokenizer,AutoTokenizer
+import torch
+import evaluate  # Bleu
+import json
+import random
+import statistics
+from sklearn.metrics import precision_score, recall_score, f1_score
+## TURN WARNINGS OFF
+import warnings
+warnings.filterwarnings("ignore")
+##13/03/23 added 
+from rouge import Rouge
+
+# Názov modelu
+DEVICE ='cuda:0'
+
+
+#T5 MODEL
+#model_name = 'T5_SK_model'
+#model_dir = "/home/omasta/T5_JUPYTER/qa_model"
+#tokenizer_dir = "/home/omasta/T5_JUPYTER/qa_tokenizer"
+
+#mT5 SMALL MODEL
+model_name = 'mT5_SMALL'
+model_dir = '/home/omasta/T5_JUPYTER/qa_model_mT5_small'
+tokenizer_dir = '/home/omasta/T5_JUPYTER/qa_tokenizer_mT5_small'
+
+#Načítanie modelu z adresára
+MODEL = T5ForConditionalGeneration.from_pretrained(model_dir, from_tf=False, return_dict=True).to(DEVICE)
+print("Model succesfully loaded!")
+TOKENIZER = AutoTokenizer.from_pretrained(tokenizer_dir, use_fast=True)
+print("Tokenizer succesfully loaded!")
+Q_LEN = 512
+TOKENIZER.add_tokens('<sep>')
+MODEL.resize_token_embeddings(len(TOKENIZER))
+
+def predict_answer(data, ref_answer=None,random=None):
+    predictions=[]
+    for i in data:
+        inputs = TOKENIZER(i['input'], max_length=Q_LEN, padding="max_length", truncation=True, add_special_tokens=True)
+        input_ids = torch.tensor(inputs["input_ids"], dtype=torch.long).to(DEVICE).unsqueeze(0)
+        attention_mask = torch.tensor(inputs["attention_mask"], dtype=torch.long).to(DEVICE).unsqueeze(0)
+        outputs = MODEL.generate(input_ids=input_ids, attention_mask=attention_mask)
+        predicted_answer = TOKENIZER.decode(outputs.flatten(), skip_special_tokens=True)
+        ref_answer = i['answer'].lower()
+        #print(ref_answer)
+        if ref_answer:
+            # Load the Bleu metric
+            bleu = evaluate.load("google_bleu")
+            #print('debug')
+            #precision = list(precision_score(ref_answer, predicted_answer))
+            #recall = list(recall_score(ref_answer, predicted_answer))
+            #f1 = list(f1_score(ref_answer, predicted_answer))
+            score = bleu.compute(predictions=[predicted_answer],
+                           	 references=[ref_answer])
+            predictions.append({'prediction':predicted_answer,'ref_answer':ref_answer,'score':score['google_bleu']})
+    return predictions
+
+def prepare_data(data):
+    articles = []
+    for article in data["data"]:
+        for paragraph in article["paragraphs"]:
+            for qa in paragraph["qas"]:
+                question = qa["question"]
+                answer = qa["answers"][0]["text"]
+                inputs = {"input": paragraph["context"]+ "<sep>" + question, "answer": answer}
+                articles.append(inputs)
+
+    return articles
+
+dev_data_path = '/home/omasta/T5_JUPYTER/skquad-221017/dev-v1.json'
+with open(dev_data_path,'r') as f:
+	data=json.load(f)
+#print('data imported')
+
+dev_data = prepare_data(data)
+
+#print('data prepared')
+print(f'Number of dev samples {len(dev_data)}')
+print(dev_data[0])
+bleu_score = []
+precisions=[]
+f1_scores=[]
+recall_scores=[]
+rouge_1 = []
+rouge_2 = []
+#X = 150
+evaluate = predict_answer(dev_data)
+rouge = Rouge()
+for item in evaluate:
+    bleu_score.append(item['score'])
+    try:
+        #scores = rouge.get_scores(item['prediction'], item['ref_answer'], avg=True)
+        precision=precision_score(list(item['ref_answer']), list(item['prediction']),average='macro')
+        recall=recall_score(list(item['ref_answer']), list(item['prediction']),average='macro')
+        f1=f1_score(list(item['ref_answer']), list(item['prediction']),average='macro')
+    except ValueError:
+        precision=0
+        recall=0
+        f1=0
+    precisions.append(precision)
+    f1_scores.append(f1)
+    recall_scores.append(recall)
+
+
+def rouge_eval(dict_x):
+    rouge = Rouge()
+    rouge_scores=[]
+    for item in dict_x:
+        if item['prediction'] and item['ref_answer']:
+            rouge_score =  rouge.get_scores(item['prediction'], item['ref_answer'])
+            rouge_scores.append(rouge_score)
+        else:
+            continue
+        return rouge_scores
+
+
+print(f'VYHODNOTENIE  VYSLEDKOV : ------------------------')
+#print(evaluate)
+#bleu_score_total = statistics.mean(bleu_score)
+#recall_score_total= statistics.mean(recall_scores)
+#f1_score_total = statistics.mean(f1_scores)
+#precision_total = statistics.mean(precisions)
+#print(f'Bleu_score of model {model_name} : ',bleu_score_total)
+#print(f'Recall of model {model_name}: ',recall_score_total)
+#print(f'F1 of model {model_name} : ', f1_score_total)
+#print(f'Precision of model {model_name}: :',precision_total)
+#print(rouge_eval(evaluate))
+print(f'{model_name} results')
+rouge_scores = rouge_eval(evaluate)
+rouge_values = [score[0]['rouge-1']['f'] for score in rouge_scores]
+mean_rouge_score = statistics.mean(rouge_values)
+print(f'Rouge:{mean_rouge_score}')
+
+rouge2_values = [score[0]['rouge-2']['f'] for score in rouge_scores]
+mean_rouge_score =statistics.mean(rouge2_values)
+print(f'Rouge-2:{mean_rouge_score}')
+