Large Language Models (LLMs) have shown remarkable performance across various tasks, yet significant disparities remain for non-English languages, and especially native African languages. This paper addresses these disparities by creating approximately 1 million human-translated words of new benchmark data in 8 low-resource African languages, covering a population of over 160 million speakers of: Amharic, Bambara, Igbo, Sepedi (Northern Sotho), Shona, Sesotho (Southern Sotho), Setswana, and Tsonga. Our benchmarks are translations of Winogrande and three sections of MMLU: college medicine, clinical knowledge, and virology. Using the translated benchmarks, we report previously unknown performance gaps between state-of-the-art (SOTA) LLMs in English and African languages. Finally, using results from over 400 fine-tuned models, we explore several methods to reduce the LLM performance gap, including high-quality dataset fine-tuning (using an LLM-as-an-Annotator), cross-lingual transfer, and cultural appropriateness adjustments. Key findings include average mono-lingual improvements of 5.6% with fine-tuning (with 5.4% average mono-lingual improvements when using high-quality data over low-quality data), 2.9% average gains from cross-lingual transfer, and a 3.0% out-of-the-box performance boost on culturally appropriate questions. The publicly available benchmarks, translations, and code from this study support further research and development aimed at creating more inclusive and effective language technologies.
Maybe not. We identify and analyse errors in the popular Massive Multitask Language Understanding (MMLU) benchmark. Even though MMLU is widely adopted, our analysis demonstrates numerous ground truth errors that obscure the true capabilities of LLMs. For example, we find that 57% of the analysed questions in the Virology subset contain errors. To address this issue, we introduce a comprehensive framework for identifying dataset errors using a novel error annotation protocol. Then, we create MMLU-Redux, which is a subset of 5,700 manually re-annotated questions across all 57 MMLU subjects. We estimate that 6.49% of MMLU questions contain errors. Using MMLU-Redux, we demonstrate significant discrepancies with the model performance metrics that were originally reported. Our results strongly advocate for revising MMLU's error-ridden questions to enhance its future utility and reliability as a benchmark. https://huggingface.co/datasets/edinburgh-dawg/mmlu-redux-2.0.
Cancer progression and monotonic accumulation models were developed to discover dependencies in the irreversible acquisition of binary traits from cross-sectional data. They have been used in computational oncology and virology but also in widely different problems such as malaria progression. These methods have been applied to predict future states of the system, identify routes of feature acquisition, and improve patient stratification, and they hold promise for evolutionary-based treatments. New methods continue to be developed. But these methods have shortcomings, which are yet to be systematically critiqued, regarding key evolutionary assumptions and interpretations. After an overview of the available methods, we focus on why inferences might not be about the processes we intend. Using fitness landscapes, we highlight difficulties that arise from bulk sequencing and reciprocal sign epistasis, from conflating lines of descent, path of the maximum, and mutational profiles, and from ambiguous use of the idea of exclusivity. We examine how the previous concerns change when bulk sequencing is explicitly considered, and underline opportunities for addressing dependencies due to frequency-dependent selection. This review identifies major standing issues, and should encourage the use of these methods in other areas with a better alignment between entities and model assumptions.
Raman spectroscopy provides a vibrational profile of the molecules and thus can be used to uniquely identify different kind of materials. This sort of fingerprinting molecules has thus led to widespread application of Raman spectrum in various fields like medical dignostics, forensics, mineralogy, bacteriology and virology etc. Despite the recent rise in Raman spectra data volume, there has not been any significant effort in developing generalized machine learning methods for Raman spectra analysis. We examine, experiment and evaluate existing methods and conjecture that neither current sequential models nor traditional machine learning models are satisfactorily sufficient to analyze Raman spectra. Both has their perks and pitfalls, therefore we attempt to mix the best of both worlds and propose a novel network architecture RamanNet. RamanNet is immune to invariance property in CNN and at the same time better than traditional machine learning models for the inclusion of sparse connectivity. Our experiments on 4 public datasets demonstrate superior performance over the much complex state-of-the-art methods and thus RamanNet has the potential to become the defacto standard in Raman spectra data analysis