File size: 11,773 Bytes
24f4359 2328b9e 24f4359 b1035cf 24f4359 b1035cf 24f4359 b1035cf 24f4359 b1035cf 24f4359 d538c84 24f4359 98c4584 24f4359 444939e 24f4359 48c15a5 24f4359 444939e 24f4359 444939e 24f4359 444939e 24f4359 444939e 9ec9054 3c14459 24f4359 444939e 4736d3c 444939e 24f4359 eabd984 24f4359 98c4584 7e9b032 24f4359 444939e 39282c7 444939e b1035cf 444939e 39282c7 24f4359 eb32620 eabd984 eb32620 24f4359 2d6b65e c00e618 2d6b65e 24f4359 4736d3c 24f4359 4736d3c 24f4359 4736d3c 24f4359 39282c7 24f4359 39282c7 24f4359 98c4584 24f4359 39282c7 36f7a59 7e9b032 36f7a59 24f4359 98c4584 24f4359 39282c7 b1035cf 39282c7 24f4359 39282c7 24f4359 4dce67f 24f4359 39282c7 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 |
---
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: "Astronomia é uma ciência natural que estuda"
example_title: Exemplo
- text: "Em um achado chocante, o cientista descobriu um"
example_title: Exemplo
- text: "Python é uma linguagem de"
example_title: Exemplo
- text: "O Gato de Botas é conhecido por"
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-162m
<img src="./logo-round.png" alt="A little llama wearing a mushroom hat and a monocle." height="200">
## 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 on Portuguese._
TeenyTinyLlama is a compact language model based on the Llama 2 architecture ([TinyLlama implementation](https://huggingface.co/TinyLlama)). This model is designed to deliver efficient natural language processing capabilities while being resource-conscious.
Also, these models were trained by leveraging [scaling laws](https://arxiv.org/abs/2203.15556) to determine the optimal number of tokens per parameter while incorporating [preference pre-training](https://arxiv.org/abs/2112.00861).
## 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](https://huggingface.co/datasets/nicholasKluge/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](https://github.com/Nkluge-correa/Aira) used to train this model. The main libraries used are:
- [Transformers](https://github.com/huggingface/transformers)
- [PyTorch](https://github.com/pytorch/pytorch)
- [Datasets](https://github.com/huggingface/datasets)
- [Tokenizers](https://github.com/huggingface/tokenizers)
- [Accelerate](https://github.com/huggingface/accelerate)
- [Codecarbon](https://github.com/mlco2/codecarbon)
## 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 ratio | 0.01 |
| gradient checkpointing | false |
| seed | 42 |
| mixed precision | 'no' |
| torch dtype | "float32" |
| tf32 | true |
## Basic usage
Using the `pipeline`:
```python
from transformers import pipeline
generator = pipeline("text-generation", model="nicholasKluge/Teeny-tiny-llama-162m")
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`:
```python
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch
# Load model and the tokenizer
tokenizer = AutoTokenizer.from_pretrained("nicholasKluge/Teeny-tiny-llama-162m", revision='main')
model = AutoModelForCausalLM.from_pretrained("nicholasKluge/Teeny-tiny-llama-162m", 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](https://arxiv.org/abs/1803.05457) | [Hellaswag](https://arxiv.org/abs/1905.07830) | [MMLU](https://arxiv.org/abs/2009.03300) | [TruthfulQA](https://arxiv.org/abs/2109.07958) |
|-------------------------------------------------------------------------------------|---------|-----------------------------------------|-----------------------------------------------|------------------------------------------|------------------------------------------------|
| [TeenyTinyLlama-162m](https://huggingface.co/nicholasKluge/TeenyTinyLlama-162m) | 31.16 | 26.15 | 29.29 | 28.11 | 41.12 |
| [Pythia-160m](https://huggingface.co/EleutherAI/pythia-160m-deduped) | 31.16 | 24.06 | 31.39 | 24.86 | 44.34 |
| [OPT-125m](https://huggingface.co/facebook/opt-125m) | 30.80 | 22.87 | 31.47 | 26.02 | 42.87 |
| [Gpt2-portuguese-small](https://huggingface.co/pierreguillou/gpt2-small-portuguese) | 30.22 | 22.48 | 29.62 | 27.36 | 41.44 |
| [Gpt2-small](https://huggingface.co/gpt2) | 29.97 | 21.48 | 31.60 | 25.79 | 40.65 |
* Evaluations on benchmarks were performed using the [Language Model Evaluation Harness](https://github.com/EleutherAI/lm-evaluation-harness) (by [EleutherAI](https://www.eleuther.ai/)). Thanks to [Laiviet](https://github.com/laiviet/lm-evaluation-harness) for translating some of the tasks in the LM-Evaluation-Harness.
## Fine Tuning
| Models | [IMDB](https://huggingface.co/datasets/christykoh/imdb_pt) | [FaQuAD-NLI](https://huggingface.co/datasets/ruanchaves/faquad-nli) | [HateBr](https://huggingface.co/datasets/ruanchaves/hatebr) |
|--------------------------------------------------------------------------------------------|------------------------------------------------------------|---------------------------------------------------------------------|-------------------------------------------------------------|
| [Teeny Tiny Llama 162m](https://huggingface.co/nicholasKluge/TeenyTinyLlama-162m) | 91.14 | 90.00 | 90.71 |
| [Bert-base-portuguese-cased](https://huggingface.co/neuralmind/bert-base-portuguese-cased) | 92.22 | 93.07 | 91.28 |
| [Gpt2-small-portuguese](https://huggingface.co/pierreguillou/gpt2-small-portuguese) | 91.60 | 86.46 | 87.42 |
## Cite as 🤗
```latex
@misc{nicholas22llama,
doi = {10.5281/zenodo.6989727},
url = {https://huggingface.co/nicholasKluge/TeenyTinyLlama-162m},
author = {Nicholas Kluge Corrêa},
title = {TeenyTinyLlama},
year = {2023},
publisher = {HuggingFace},
journal = {HuggingFace repository},
}
```
## License
The TeenyTinyLlama-162m is licensed under the Apache License, Version 2.0. See the [LICENSE](LICENSE) file for more details. |