Llama-Carvalho-PT-GL

Table of Contents

Click to expand

• Llama-Carvalho-PT-GL
  • Table of Contents
  • Model description
  • Intended uses and limitations
  • How to use
  • Training
    • Tools
    • Training data
    • Training hyperparameters
    • Framework
  • Evaluation
  • Additional information
    • Contact
    • License
    • Funding

Model description

Llama-Carvalho-PT-GL is a 8B-parameter transformer-based causal language model for Galician, Portuguese, Spanish and English. It is the result of a continual pretraining of meta-llama/Llama-3.1-8B with a multilingual corpus consisting of 540M tokens of plain text and 72M tokens of instructions (formated as plain text)

This model is part of the Carvalho familily, a family of LLMs specialized in Portuguese and Galician which can be found here.

Intended uses and limitations

The Llama-Carvalho-PT-GL model is ready-to-use only for causal language modeling. It can perform text-generation tasks and be fine-tuned for specific scenarios.

How to use

import torch
from transformers import pipeline, AutoTokenizer, AutoModelForCausalLM

input_text = "Hoxe fai un bo día. O sol  "

model_id  = "Nos-PT/Llama-Carvalho-PT-GL"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id)
generator = pipeline(
    "text-generation",
    model=model,
    tokenizer=tokenizer,
    torch_dtype=torch.bfloat16,
    trust_remote_code=True,
    device_map="auto",
)
generation = generator(
    input_text,
    do_sample=True,
    top_k=10,
    eos_token_id=tokenizer.eos_token_id
)

print(f"Result: {generation[0]['generated_text']}")

Training

Tools

It was trained using HuggingFace Transformers and Pytorch, using the Causal Modeling Language script. We also use DeepSpeed to deal with the huge size of the model.

Training data

The training corpus consists of texts in 4 languages, with an emphasis on Portuguese and Galician. The main aim of this is to ensure that the model learns to work with this language perfectly, while maintaining knowledge of languages already known (Spanish, English), learning others (Galician) or adapting existing language varieties (Portuguese-PT instead of Portuguese-BR).

The corpus is composed as follows:

Corpus		gl	pt	es	en
Base plain text corpus	Tokens	232M	250M	29M	29M
	Percentage (of the total base corpus)	42,96%	46,29%	5,37%	5,37%
Instructions	Tokens	26,7M	44M	804K	623K
	Percentage (of the total instructions corpus)	37,01%	61,00%	1,11%	0,86%

Training hyperparameters

• seed: 42
• num_devices: 5
• train_batch_size: 4
• eval_batch_size: 4
• gradient_acummulation: 8
• optimizer: AdamW
• betas: (0.9,0.999)
• epsilon: 1e-08
• weight_decay_rate: 0.1
• scheduler: "Linear"
• learning_rate: 1e-04
• num_epochs: 1.0

Framework

The training was conducted on the MareNostrum V in the Barcelona Supercomputing Center (BSC), using 10 nodes with 4 GPUs NVIDIA H100 64GB each one.

Evaluation

Galician and European Portuguese

Soon...

American Portuguese: Open Portuguese LLM Leaderboard Evaluation Results

Detailed results can be found here and on the 🚀 Open Portuguese LLM Leaderboard

Metric	Value
Average	60.06
ENEM Challenge (No Images)	19.38
BLUEX (No Images)	18.92
Assin2 RTE	89.30
Assin2 STS	78.18
FaQuAD NLI	65.88
HateBR Binary	82.83
PT Hate Speech Binary	64.84
tweetSentBR	61.15

Additional information

Contact

For further information, please send an email to

License

MIT License

Copyright (c) 2024

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Funding

This model was development within the Nós Project, funded by the Ministerio para la Transformación Digital y de la Función Pública - Funded by EU – NextGenerationEU within the framework of the project ILENIA with reference 2022/TL22/00215336.