Bielik 11B v2 Technical Report

TL;DR

Bielik 11B v2 is a highly efficient Polish language model with innovative training techniques that outperform larger models and set new standards for resource-efficient language processing in Polish.

Contribution

The paper introduces Weighted Instruction Cross-Entropy Loss and Adaptive Learning Rate, enhancing training effectiveness and model performance for Polish language modeling.

Findings

Outperforms larger models with fewer parameters

Surpasses other Polish language models on multiple benchmarks

Demonstrates strong cross-lingual capabilities

Abstract

We present Bielik 11B v2, a state-of-the-art language model optimized for Polish text processing. Built on the Mistral 7B v0.2 architecture and scaled to 11B parameters using depth up-scaling, this model demonstrates exceptional performance across Polish language benchmarks while maintaining strong cross-lingual capabilities. We introduce two key technical innovations: Weighted Instruction Cross-Entropy Loss, which optimizes learning across diverse instruction types by assigning quality-based weights to training examples, and Adaptive Learning Rate, which dynamically adjusts based on context length. Comprehensive evaluation across multiple benchmarks demonstrates that Bielik 11B v2 outperforms many larger models, including those with 2-6 times more parameters, and significantly surpasses other specialized Polish language models on tasks ranging from linguistic understanding to complex reasoning. The model's parameter efficiency and extensive quantization options enable deployment across various hardware configurations, advancing Polish language AI capabilities and establishing new benchmarks for resource-efficient language modeling in less-represented languages.