app.py

from transformers import Trainer, TrainingArguments, DeepSeekForQuestionAnswering, DeepSeekTokenizer
from datasets import load_dataset

# Load the DeepSeek model and tokenizer
model = DeepSeekForQuestionAnswering.from_pretrained("DeepSeek/DeepSeek-v3")
tokenizer = DeepSeekTokenizer.from_pretrained("DeepSeek/DeepSeek-v3")

# Load dataset
dataset = load_dataset("json", data_files={"train": "your_dataset_train.json", "test": "your_dataset_test.json"})

# Tokenize the dataset
def tokenize_function(examples):
    return tokenizer(examples['question'], examples['document'], truncation=True, padding=True)

tokenized_datasets = dataset.map(tokenize_function, batched=True)

# Set up the training arguments
training_args = TrainingArguments(
    output_dir='./results', 
    evaluation_strategy="epoch",
    learning_rate=2e-5,
    per_device_train_batch_size=16,
    per_device_eval_batch_size=16,
    num_train_epochs=3,
    weight_decay=0.01
)

# Initialize Trainer
trainer = Trainer(
    model=model,
    args=training_args,
    train_dataset=tokenized_datasets['train'],
    eval_dataset=tokenized_datasets['test']
)

# Start the fine-tuning
trainer.train()

# Save the model after fine-tuning
model.save_pretrained('./fine_tuned_deepseek')