metadata

license: apache-2.0
datasets:
  - nicholasKluge/portuguese-corpus-v3
language:
  - pt
metrics:
  - perplexity
library_name: transformers
pipeline_tag: text-generation
tags:
  - text-generation-inference
widget:
  - text: >-
      Mamonas Assassinas, também conhecida como Mamonas, foi uma banda
      brasileira de rock formada em Guarulhos em 1995, originada da banda
      Utopia. Notabilizaram-se por suas composições que conciliavam pop rock com
      gêneros musicais populares, tais como
    example_title: Exemplo
  - text: >-
      A adaptação cinematográfica de The Last of the Mohicans (1977) foi
      dirigida pelo cineasta americano James L. Conway. A produção foi produzida
      pela
    example_title: Exemplo
  - text: >-
      Frente parlamentar pede a Pacheco que devolva ao governo MP que revisa
      desoneração.


      Criticada por parlamentares e setores, MP foi editada nesta sexta e
      reonera, de forma gradual, a folha de pagamento de 17 segmentos. Eventual
      devolução invalidaria efeitos da medida.
    example_title: Exemplo
  - text: >-
      Salve salve Pythonista!


      Nesse post você vai aprender a função mais utilizada por todo Pythonista:
      a função print! Usamos a função print para mostrar dados na tela dos
      usuários, através do terminal. A função para imprimir dados em Python é a
      função `print()`. Ela é responsável por mostrar valores em seu terminal:
    example_title: Exemplo
inference:
  parameters:
    repetition_penalty: 1.2
    temperature: 0.2
    top_k: 20
    top_p: 0.2
    max_new_tokens: 150
co2_eq_emissions:
  emissions: 5.6
  source: CodeCarbon
  training_type: pre-training
  geographical_location: Germany
  hardware_used: NVIDIA A100-SXM4-40GB

TeenyTinyLlama-160m

Model Summary

Given the lack of available monolingual foundational models in non-English languages and the fact that some of the most used and downloaded models by the community are those small enough to allow individual researchers and hobbyists to use them in low-resource environments, we developed the TeenyTinyLlama: a series of small foundational models trained in Brazilian Portuguese.

TeenyTinyLlama is a series of compact language models based on the Llama 2 architecture. These models are designed to deliver efficient natural language processing capabilities while being resource-conscious.

Also, TeenyTinyLlama models were trained by leveraging scaling laws to determine the optimal number of tokens per parameter while incorporating preference pre-training.

Details

Architecture: a Transformer-based model pre-trained via causal language modeling
Size: 162,417,408 million parameters
Context length: 2048 tokens
Dataset: Portuguese-Corpus-v3 (6.2B tokens)
Language: Portuguese
Number of steps: 457,969 (3.7B tokens)
GPU: 1 NVIDIA A100-SXM4-40GB
Training time: ~ 36 hours
Emissions: 5.6 KgCO2 (Germany)
Total energy consumption: 15.5 kWh

This repository has the source code used to train this model. The main libraries used are:

Training Set-up

These are the main arguments used in the training of this model:

Arguments	Value
vocabulary size	32000
hidden dimension size	768
intermediate dimension size	3072
context length	2048
nº attention heads	12
nº hidden layers	12
nº key value heads	12
nº training samples	1831873
nº validation samples	18000
nº epochs	1
evaluation steps	100000
train batch size	4
eval batch size	4
gradient accumulation steps	1
optimizer	torch.optim.AdamW
learning rate	0.0006
adam epsilon	0.00000001
weight decay	0.01
scheduler type	"cosine"
warmup steps	5000
gradient checkpointing	false
seed	42
mixed precision	'no'
torch dtype	"float32"
tf32	true

Intended Uses

The primary intended use of TeenyTinyLlama is to research the behavior, functionality, and limitations of large language models. Checkpoints saved during training are intended to provide a controlled setting for performing scientific experiments. You may also further fine-tune and adapt TeenyTinyLlama-160m for deployment, as long as your use is in accordance with the Apache 2.0 license. If you decide to use pre-trained TeenyTinyLlama-160m as a basis for your fine-tuned model, please conduct your own risk and bias assessment.

Basic usage

Using the pipeline:

from transformers import pipeline

generator = pipeline("text-generation", model="nicholasKluge/Teeny-tiny-llama-160m")

completions  = generator("Astronomia é a ciência", num_return_sequences=2, max_new_tokens=100)

for comp in completions:
  print(f"🤖 {comp['generated_text']}")

Using the AutoTokenizer and AutoModelForCausalLM:

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

# Load model and the tokenizer
tokenizer = AutoTokenizer.from_pretrained("nicholasKluge/Teeny-tiny-llama-160m", revision='main')
model = AutoModelForCausalLM.from_pretrained("nicholasKluge/Teeny-tiny-llama-160m", revision='main')

# Pass the model to your device
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

model.eval()
model.to(device)

# Tokenize the inputs and pass them to the device
inputs = tokenizer("Astronomia é a ciência", return_tensors="pt").to(device)

# Generate some text
completions = model.generate(**inputs, num_return_sequences=2, max_new_tokens=100)

# Print the generated text
for i, completion in enumerate(completions):
    print(f'🤖 {tokenizer.decode(completion)}')

Limitations

Hallucinations: This model can produce content that can be mistaken for truth but is, in fact, misleading or entirely false, i.e., hallucination.
Biases and Toxicity: This model inherits the social and historical stereotypes from the data used to train it. Given these biases, the model can produce toxic content, i.e., harmful, offensive, or detrimental to individuals, groups, or communities.
Unreliable Code: The model may produce incorrect code snippets and statements. These code generations should not be treated as suggestions or accurate solutions.
Language Limitations: The model is primarily designed to understand standard Portuguese (BR). Other languages might challenge its comprehension, leading to potential misinterpretations or errors in response.
Repetition and Verbosity: The model may get stuck on repetition loops (especially if the repetition penalty during generations is set to a meager value) or produce verbose responses unrelated to the prompt it was given.

Evaluations

Steps	Evaluation Loss	Perplexity	Total Energy Consumption	Emissions
100.000	3.19	24.52	3.75 kWh	1.28 CO2eq
200.000	3.02	20.58	7.51 kWh	2.56 CO2eq
300.000	2.83	16.98	11.25 kWh	3.84 CO2eq
400.000	2.79	16.41	14.52 kWh	5.11 CO2eq

Benchmarks

Models	Average	ARC	Hellaswag	MMLU	TruthfulQA
TeenyTinyLlama-160m	31.16	26.15	29.29	28.11	41.12
Pythia-160m*	31.16	24.06	31.39	24.86	44.34
OPT-125m*	30.80	22.87	31.47	26.02	42.87
Gpt2-portuguese-small	30.22	22.48	29.62	27.36	41.44
Gpt2-small8	29.97	21.48	31.60	25.79	40.65
Xglm-564M*	31.20	24.57	34.64	25.18	40.43
Bloom-560m*	32.13	24.74	37.15	24.22	42.44

Evaluations on benchmarks were performed using the Language Model Evaluation Harness (by EleutherAI). Thanks to Laiviet for translating some of the tasks in the LM-Evaluation-Harness. The results of models marked with an "*" were retirved from the Open LLM Leaderboard.

Fine-Tuning Comparisons

Models	IMDB	FaQuAD-NLI	HateBr	Assin2	AgNews
Teeny Tiny Llama 160m	91.14	90.00	90.71	85.78	94.05
Bert-base-portuguese-cased	92.22	93.07	91.28	87.45	94.19
Bert-large-portuguese-cased	93.58	92.26	91.57	88.97	94.11
Gpt2-small-portuguese	91.60	86.46	87.42	86.11	94.07

Cite as 🤗


@misc{nicholas22llama,
  doi = {10.5281/zenodo.6989727},
  url = {https://huggingface.co/nicholasKluge/TeenyTinyLlama-160m},
  author = {Nicholas Kluge Corrêa},
  title = {TeenyTinyLlama},
  year = {2023},
  publisher = {HuggingFace},
  journal = {HuggingFace repository},
}

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-160m is licensed under the Apache License, Version 2.0. See the LICENSE file for more details.