Top LLM Papers of the week (February 2024 Week 4)

[1] The Era of 1-bit LLMs-All Large Language Models are in 1.58 Bits (link)

Introduces BitNet b1.58, a 1-bit LLM that achieves performance on par with full-precision LLMs and also reduces computational costs significantly.

[2] MobiLlama: Towards Accurate and Lightweight Fully Transparent GPT (link)

Introduces MobiLlama (0.5B parameters), a smaller language model (SLM) developed for privacy-sensitive and resource-constrained applications.

[3] CodeS: Towards Building Open-source Language Models for Text-to-SQL (link)

Introduces CodeS, an open-source LLM family models (1B to 15B sizes) developed for text-to-SQL translation task.

[4] A Survey on Data Selection for Language Models (link)

Provides a comprehensive survey of data selection methods to train LLMs.

[5] MobileLLM: Optimizing Sub-billion Parameter Language Models for On-Device Use Cases (link)

Introduces the MobileLLM family of language models (125M and 350M parameters), development for efficient on-device deployment.

[6] A First Look at GPT Apps: Landscape and Vulnerability (link)

Investigates security and plagiarism issues within the GPT store introduced by OpenAI. Findings showed vulnerabilities which lead to easy access to internal code and subsequently duplication of GPTs.

[7] Large Language Models on Tabular Data - A Survey (link)

Provides a much-needed comprehensive survey of applications of large language models (LLMs) for various tasks on tabular data.

[8] Learning to Generate Instruction Tuning Datasets for Zero-Shot Task Adaptation (link)

Introduces Bonito, an open-source model that generates instruction-tuning dataset instances from unlabeled text data.

[9] Large Multimodal Agents: A Survey (link)

Provides a systematic survey of LLM-driven multimodal agents, which expand the capabilities of language models by incorporating multimodal inputs.

[10] tinyBenchmarks: evaluating LLMs with fewer examples (link)

The authors show that significantly smaller subsets of popular benchmarks can be used for accurate performance assessment.

If you like this, do subscribe to the newsletter so that you won't miss reading interesting LLM papers.

Are you are interested in learning LLM Prompting Engineering, here is an excellent book (free and available online to read)

Book link: https://github.com/AkmmusAI/LLM-Prompt-Engineering-Simplified-Book

Enjoy learning and using LLMs. See you in the next week with another set of interesting LLM papers .