TeenyTinyLlama-460m-AgNews

TeenyTinyLlama is a pair of small foundational models trained in Brazilian Portuguese.

This repository contains a version of TeenyTinyLlama-460m (TeenyTinyLlama-460m-AgNews) fine-tuned on the AgNews dataset.

Details

• Number of Epochs: 3
• Batch size: 16
• Optimizer: torch.optim.AdamW (learning_rate = 4e-5, epsilon = 1e-8)
• GPU: 1 NVIDIA A100-SXM4-40GB

Usage

Using transformers.pipeline:

from transformers import pipeline

text = "Quando falamos de Inteligência Artificial, hoje em dia existem já vários sistemas que começam a ganhar popularidade,\
 embora nenhum seja ainda tão conhecido como o ChatGPT da OpenAI."

classifier = pipeline("text-classification", model="nicholasKluge/TeenyTinyLlama-460m-AgNews")
classifier(text)

# >>> [{'label': 'TECNOLOGIA', 'score': 0.9997298121452332}] 

Reproducing

To reproduce the fine-tuning process, use the following code snippet:

# AgNews
!pip install transformers datasets evaluate accelerate -q

import evaluate
import numpy as np
from datasets import load_dataset, Dataset, DatasetDict
from transformers import AutoTokenizer, DataCollatorWithPadding
from transformers import AutoModelForSequenceClassification, TrainingArguments, Trainer

# Load the task
dataset = load_dataset("maritaca-ai/ag_news_pt")

# Create a `ModelForSequenceClassification`
model = AutoModelForSequenceClassification.from_pretrained(
    "nicholasKluge/TeenyTinyLlama-460m",
    num_labels=4,
    id2label={0: "MUNDO", 1: "ESPORTES", 2: "NEGÓCIOS", 3: "TECNOLOGIA"},
    label2id={"MUNDO": 0, "ESPORTES": 1, "NEGÓCIOS": 2, "TECNOLOGIA": 3}
)

tokenizer = AutoTokenizer.from_pretrained("nicholasKluge/TeenyTinyLlama-460m")

# Format the dataset
train = dataset['train'].to_pandas()
train["text"] = train["title"] + "\n\n" + train["text"]
train.label = train.label.astype(int)
train = Dataset.from_pandas(train)

test = dataset['test'].to_pandas()
test["text"] = test["title"] + "\n\n" + test["text"]
test.label = test.label.astype(int)
test = Dataset.from_pandas(test)

dataset = DatasetDict({
    "train": train,
    "test": test
})

# Preprocess the dataset
def preprocess_function(examples):
    return tokenizer(examples["text"], truncation=True, max_length=256)

dataset_tokenized = dataset.map(preprocess_function, batched=True)

# Create a simple data collactor
data_collator = DataCollatorWithPadding(tokenizer=tokenizer)

# Use accuracy as an evaluation metric
accuracy = evaluate.load("accuracy")

# Function to compute accuracy
def compute_metrics(eval_pred):
    predictions, labels = eval_pred
    predictions = np.argmax(predictions, axis=1)
    return accuracy.compute(predictions=predictions, references=labels)

# Define training arguments
training_args = TrainingArguments(
    output_dir="checkpoints",
    learning_rate=4e-5,
    per_device_train_batch_size=16,
    per_device_eval_batch_size=16,
    num_train_epochs=3,
    weight_decay=0.01,
    evaluation_strategy="epoch",
    save_strategy="epoch",
    load_best_model_at_end=True,
    push_to_hub=True,
    hub_token="your_token_here",
    hub_model_id="username/model-ID",
)

# Define the Trainer
trainer = Trainer(
    model=model,
    args=training_args,
    train_dataset=dataset_tokenized["train"],
    eval_dataset=dataset_tokenized["test"],
    tokenizer=tokenizer,
    data_collator=data_collator,
    compute_metrics=compute_metrics,
)

# Train!
trainer.train()

Fine-Tuning Comparisons

To further evaluate the downstream capabilities of our models, we decided to employ a basic fine-tuning procedure for our TTL pair on a subset of tasks from the Poeta benchmark. We apply the same procedure for comparison purposes on both BERTimbau models, given that they are also LLM trained from scratch in Brazilian Portuguese and have a similar size range to our models. We used these comparisons to assess if our pre-training runs produced LLM capable of producing good results ("good" here means "close to BERTimbau") when utilized for downstream applications.

Models	IMDB	FaQuAD-NLI	HateBr	Assin2	AgNews	Average
BERTimbau-large	93.58	92.26	91.57	88.97	94.11	92.10
BERTimbau-small	92.22	93.07	91.28	87.45	94.19	91.64
TTL-460m	91.64	91.18	92.28	86.43	94.42	91.19
TTL-160m	91.14	90.00	90.71	85.78	94.05	90.34

All the shown results are the higher accuracy scores achieved on the respective task test sets after fine-tuning the models on the training sets. All fine-tuning runs used the same hyperparameters, and the code implementation can be found in the model cards of our fine-tuned models.

Cite as 🤗

@misc{correa24ttllama,
  title = {TeenyTinyLlama: open-source tiny language models trained in Brazilian Portuguese},
  author = {Corr{\^e}a, Nicholas Kluge and Falk, Sophia and Fatimah, Shiza and Sen, Aniket and De Oliveira, Nythamar},
  journal={arXiv preprint arXiv:2401.16640},
  year={2024}
}

@misc{correa24ttllama,
  doi = {10.1016/j.mlwa.2024.100558},
  url = {https://www.sciencedirect.com/science/article/pii/S2666827024000343},
  title = {TeenyTinyLlama: open-source tiny language models trained in Brazilian Portuguese},
  author = {Corr{\^e}a, Nicholas Kluge and Falk, Sophia and Fatimah, Shiza and Sen, Aniket and De Oliveira, Nythamar},
  journal={Machine Learning With Applications},
  publisher = {Springer},
  year={2024}
}

Funding

This repository was built as part of the RAIES (Rede de Inteligência Artificial Ética e Segura) initiative, a project supported by FAPERGS - (Fundação de Amparo à Pesquisa do Estado do Rio Grande do Sul), Brazil.

License

TeenyTinyLlama-460m-AgNews is licensed under the Apache License, Version 2.0. See the LICENSE file for more details.