README.md

---
license: apache-2.0
datasets:
- ruanchaves/faquad-nli
language:
- pt
metrics:
- accuracy
library_name: transformers
pipeline_tag: text-classification
tags:
- textual-entailment
---
# TeenyTinyLlama-162m-FAQUAD

TeenyTinyLlama is a series of small foundational models trained on Portuguese.

This repository contains a version of [TeenyTinyLlama-162m](https://huggingface.co/nicholasKluge/TeenyTinyLlama-162m) fine-tuned on the [FAQUAD dataset]().

## Reproducing

```python
# Faquad-nli
! pip install transformers datasets evaluate accelerate -q

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

# Basic fine-tuning arguments
token="your_token"
task="ruanchaves/faquad-nli"
model_name="nicholasKluge/Teeny-tiny-llama-162m"
output_dir="checkpoint"
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"
hub_model_id="nicholasKluge/Teeny-tiny-llama-162m-faquad"

# Login on the hub to load and push
login(token=token)

# Load the task
dataset = load_dataset(task)

# Create a `ModelForSequenceClassification`
model = AutoModelForSequenceClassification.from_pretrained(
    model_name, 
    num_labels=2, 
    id2label={0: "UNSUITABLE", 1: "SUITABLE"}, 
    label2id={"UNSUITABLE": 0, "SUITABLE": 1}
)

tokenizer = AutoTokenizer.from_pretrained(model_name)

# If model does not have a pad_token, we need to add it
#tokenizer.pad_token = tokenizer._eos_token
#model.config.pad_token_id = model.config.eos_token_id

# Preprocess if needed
train = dataset['train'].to_pandas()
train['text'] = train['question'] + tokenizer.bos_token + train['answer'] + tokenizer.eos_token
train = train[['text', 'label']]
train.labels = train.label.astype(int)
train = Dataset.from_pandas(train)

test = dataset['test'].to_pandas()
test['text'] = test['question'] + tokenizer.bos_token + test['answer'] + tokenizer.eos_token
test = test[['text', 'label']]
test.labels = test.label.astype(int)
test = Dataset.from_pandas(test)

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

# Pre process the dataset
def preprocess_function(examples):
    return tokenizer(examples["text"], truncation=True)

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

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

# Use accuracy as 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=output_dir,
    learning_rate=learning_rate,
    per_device_train_batch_size=per_device_train_batch_size,
    per_device_eval_batch_size=per_device_eval_batch_size,
    num_train_epochs=num_train_epochs,
    weight_decay=weight_decay,
    evaluation_strategy=evaluation_strategy,
    save_strategy=save_strategy,
    load_best_model_at_end=True,
    push_to_hub=True,
    hub_token=token,
    hub_private_repo=True,
    hub_model_id=hub_model_id,
    tf32=True,
)

# 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()

```