fine_tune_from_checkpoint.py

from transformers import AutoModelForSequenceClassification, AutoTokenizer, Trainer, TrainingArguments
from datasets import load_dataset
import numpy as np
from sklearn.metrics import accuracy_score, precision_recall_fscore_support

# Load the previously saved model and tokenizer
model_path = "./further_fine_tuned_model"
model = AutoModelForSequenceClassification.from_pretrained(model_path)
tokenizer = AutoTokenizer.from_pretrained(model_path)

# Load and prepare dataset
dataset = load_dataset("imdb")
train_dataset = dataset["train"].shuffle(seed=42).select(range(2000))
eval_dataset = dataset["test"].shuffle(seed=42).select(range(200))

def tokenize_function(examples):
    return tokenizer(examples["text"], padding="max_length", truncation=True, max_length=512)

# Tokenize the datasets
train_dataset = train_dataset.map(tokenize_function, batched=True, remove_columns=["text"])
eval_dataset = eval_dataset.map(tokenize_function, batched=True, remove_columns=["text"])

# Set data types
train_dataset.set_format(type='torch', columns=['input_ids', 'attention_mask', 'label'])
eval_dataset.set_format(type='torch', columns=['input_ids', 'attention_mask', 'label'])

def compute_metrics(eval_pred):
    logits, labels = eval_pred
    predictions = np.argmax(logits, axis=-1)
    precision, recall, f1, _ = precision_recall_fscore_support(labels, predictions, average='binary')
    acc = accuracy_score(labels, predictions)
    return {
        'accuracy': acc,
        'f1': f1,
        'precision': precision,
        'recall': recall
    }

# Set up training arguments
training_args = TrainingArguments(
    output_dir="./continued_training_results",
    num_train_epochs=3,  # You can adjust this
    per_device_train_batch_size=8,
    per_device_eval_batch_size=16,
    warmup_steps=500,
    weight_decay=0.01,
    logging_dir="./continued_training_logs",
    evaluation_strategy="epoch",
    save_strategy="epoch",
    load_best_model_at_end=True,
    metric_for_best_model="accuracy",
    no_cuda=True,
)

# Initialize the Trainer
trainer = Trainer(
    model=model,
    args=training_args,
    train_dataset=train_dataset,
    eval_dataset=eval_dataset,
    compute_metrics=compute_metrics,
)

# Continue training
trainer.train()

# Save the further fine-tuned model
output_dir = "./further_fine_tuned_model2"
trainer.save_model(output_dir)
tokenizer.save_pretrained(output_dir)

print("Further model training completed successfully.")