---
license: apache-2.0
datasets:
- csebuetnlp/xlsum
- SGaleshchuk/XL_SUM_ukr_synthetic_hallucinations
language:
- uk
- en
tags:
- text-generation-inference
pipeline_tag: text-generation
---

The following training arguments used for Llama-2 finetuning with Ukrainian corpora pf XL-SUM: 
- learning-rate=2e-4,
- maximum number of tokens=512,
- 15 epochs.
  Lora perf arguments:
- rank = 32,
- lora-alpha=16,
- dropout = 0.1.