Falcon2-11B Technical Report

TL;DR

Falcon2-11B and its multimodal version Falcon2-11B-vlm are large foundation models trained on extensive datasets, demonstrating strong performance across various benchmarks and tasks, with insights into training strategies and effects.

Contribution

Introduces Falcon2-11B and Falcon2-11B-vlm models, detailing their training process, evaluation, and open-sourcing, advancing large-scale foundation and multimodal models.

Findings

Strong generalization across benchmarks

Higher average scores than similar open-source models

Training insights on batch size and learning rate effects

Abstract

We introduce Falcon2-11B, a foundation model trained on over five trillion tokens, and its multimodal counterpart, Falcon2-11B-vlm, which is a vision-to-text model. We report our findings during the training of the Falcon2-11B which follows a multi-stage approach where the early stages are distinguished by their context length and a final stage where we use a curated, high-quality dataset. Additionally, we report the effect of doubling the batch size mid-training and how training loss spikes are affected by the learning rate. The downstream performance of the foundation model is evaluated on established benchmarks, including multilingual and code datasets. The foundation model shows strong generalization across all the tasks which makes it suitable for downstream finetuning use cases. For the vision language model, we report the performance on several benchmarks and show that our model achieves a higher average score compared to open-source models of similar size. The model weights and code of both Falcon2-11B and Falcon2-11B-vlm are made available under a permissive license.