Model Card for Minerva-350M-base-v1.0

Minerva is the first family of LLMs pretrained from scratch on Italian developed by Sapienza NLP in collaboration with Future Artificial Intelligence Research (FAIR) and CINECA. Notably, the Minerva models are truly-open (data and model) Italian-English LLMs, with approximately half of the pretraining data including Italian text.

• Minerva LLMs - website

Description

This is the model card for Minerva-350M-base-v1.0, a 350 million parameter model trained on 70 billion tokens (35 billion in Italian, 35 billion in English).

This model is part of the Minerva LLM family:

• Minerva-350M-base-v1.0
• Minerva-1B-base-v1.0
• Minerva-3B-base-v1.0

How to use Minerva with Hugging Face transformers

import transformers
import torch

model_id = "sapienzanlp/Minerva-350M-base-v1.0"

# Initialize the pipeline.
pipeline = transformers.pipeline(
    "text-generation",
    model=model_id,
    model_kwargs={"torch_dtype": torch.bfloat16},
    device_map="auto",
)

# Input text for the model.
input_text = "La capitale dell'Italia è"

# Compute the outputs.
output = pipeline(
  input_text,
  max_new_tokens=128,
)

# Output:
# [{'generated_text': "La capitale dell'Italia è la città di Roma, che si trova a [...]"}]

Model Architecture

Minerva-350M-base-v1.0 is a Transformer model based on the Mistral architecture, where the number of layers, number of heads, and the hidden states dimension are modified to reach 350M parameters. Please, take a look at the configuration file for a detailed breakdown of the hyperparameters we chose for this model.

The Minerva LLM family is composed of:

Model Name	Tokens	Layers	Hidden Size	Attention Heads	KV Heads	Sliding Window	Max Context Length
Minerva-350M-base-v1.0	70B (35B it + 35B en)	16	1152	16	4	2048	16384
Minerva-1B-base-v1.0	200B (100B it + 100B en)	16	2048	16	4	2048	16384
Minerva-3B-base-v1.0	660B (330B it + 330B en)	32	2560	32	8	2048	16384

Model Training

Minerva-350M-base-v1.0 was trained using llm-foundry 0.6.0 from MosaicML. The hyperparameters used are the following:

Model Name	Optimizer	lr	betas	eps	weight decay	Scheduler	Warmup Steps	Batch Size (Tokens)	Total Steps
Minerva-350M-base-v1.0	Decoupled AdamW	2e-4	(0.9, 0.95)	1e-8	0.0	Cosine	2%	4M	16,690
Minerva-1B-base-v1.0	Decoupled AdamW	2e-4	(0.9, 0.95)	1e-8	0.0	Cosine	2%	4M	47,684
Minerva-3B-base-v1.0	Decoupled AdamW	2e-4	(0.9, 0.95)	1e-8	0.0	Cosine	2%	4M	157,357

Model Evaluation

We assessed our model using the LM-Evaluation-Harness library, which serves as a comprehensive framework for testing generative language models across a wide range of evaluation tasks.

All the reported benchmark data was already present in the LM-Evaluation-Harness suite.

Italian Data:

Task	Accuracy
xcopa (0-shot)	0.554
Hellaswag (5-shot)	0.3205
Belebele (5-shot)	0.2489
TruthfulQA MC 1 (0-shot)	0.2503
TruthfulQA MC 2 (0-shot)	0.4080
M MMLU (5-shot)	0.2544
arc challenge (5-shot)	0.2293

English Data:

Task	Accuracy
Hellaswag (5-shot)	0.3248
Belebele (5-shot)	0.2200
TruthfulQA MC 1 (0-shot)	0.2497
TruthfulQA MC 2 (0-shot)	0.4438
M MMLU (5-shot)	0.2560
arc challenge (5-shot)	0.2372
arc easy (5-shot)	0.3965
piqa (5-shot)	0.6540
sciq (5-shot)	0.6730

Training Data

Minerva-350M-base-v1.0 was trained on 35B Italian tokens and 35B English tokens sampled from CulturaX. The data was selected from the following sources:

• OSCAR-2201
• OSCAR-2301
• mC4

We have extracted some statistics on Italian (115B tokens) and English (210B tokens) documents from CulturaX on the selected sources:

Proportion of number of tokens per domain (Italian)

Proportion of number of tokens per domain (English)

Tokenizer Fertility

The tokenizer fertility measures the average amount of tokens produced per tokenized word. A tokenizer displaying high fertility values in a particular language typically indicates that it segments words in that language extensively. The tokenizer fertility is strictly correlated with the inference speed of the model with respect to a specific language, as higher values mean longer sequences of tokens to generate and thus lower inference speed.

Fertility computed over a sample of Cultura X (CX) data and Wikipedia (Wp):

Model	Voc. Size	Fertility IT (CX)	Fertility EN (CX)	Fertility IT (Wp)	Fertility EN (Wp)
Mistral-7B-v0.1	32000	1.87	1.32	2.05	1.57
gemma-7b	256000	1.42	1.18	1.56	1.34
Minerva-350M-base-v1.0	32768	1.39	1.32	1.66	1.59

Notice

Minerva-350M-base-v1.0 is a pretrained base model and, therefore, has no moderation mechanisms.

The Sapienza NLP Team

• Riccardo Orlando: data preprocessing, model training
• Pere-Lluis Huguet Cabot: data preprocessing, vocabulary, evaluation
• Luca Moroni: data curation, data analysis, downstream tasks, evaluation
• Simone Conia: data curation, evaluation, project supervision
• Edoardo Barba: data preprocessing, downstream tasks, project supervision
• Roberto Navigli: project coordinator

Special thanks for their support

• Giuseppe Fiameni, Nvidia
• Sergio Orlandini, CINECA

Acknowledgments

This work was funded by the PNRR MUR project PE0000013-FAIR. We acknowledge the CINECA award "IscB_medit" under the ISCRA initiative, for the availability of high performance computing resources and support.