Artificial Intelligence

Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer

# Summary This paper introduces T5 (Text-to-Text Transfer Transformer), a unified framework for text-based tasks. The authors propose converting all NLP tasks into a text-to-text format, where the inp...

Scaling Laws for Neural Language Models

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper, published by OpenAI, investigates the relationship between the performance of neural language models (NLMs) and three key factors: model s...

Evaluating Large Language Models Trained on Code

<h2 class="text-2xl font-bold mb-4">Summary</h2> This OpenAI paper, published in August 2021, focuses on evaluating large language models (LLMs) specifically trained on code. It likely introduces and...

On the Opportunities and Risks of Foundation Models

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper, published by Stanford in August 2021, examines the landscape of foundation models – models trained on broad data that can be adapted to a ...

Finetuned Language Models are Zero-Shot Learners

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper from Google introduces the concept of Finetuned Language Models Are Zero-Shot Learners (FLAN). It investigates the zero-shot capabilities o...

Multitask Prompted Training Enables Zero-Shot Task Generalization

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper introduces a novel approach to zero-shot task generalization using multitask prompted training. The authors propose a method that leverages...

WebGPT: Browser-assisted question-answering with human feedback

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper introduces WebGPT, a question-answering system designed to answer questions using the internet, specifically with browser assistance. It le...

Scaling Language Models: Methods, Analysis & Insights from Training Gopher

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper details the training and analysis of Gopher, a large language model (LLM) developed by DeepMind. It explores the scaling laws of LLMs, inve...

Chain-of-Thought Prompting Elicits Reasoning in Large Language Models

<h2 class="text-2xl font-bold mb-4">Summary</h2> This research paper, published in 2022 by Google, introduces and investigates the effectiveness of "Chain-of-Thought" (COT) prompting for eliciting re...

LaMDA: Language Models for Dialog Applications

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper introduces LaMDA (Language Model for Dialog Applications), a large language model specifically designed for engaging and informative dialog...

Solving Quantitative Reasoning Problems with Language Models

<h2 class="text-2xl font-bold mb-4">Summary</h2> This research paper, published by Google in January 2022, focuses on solving quantitative reasoning problems using Language Models (LMs). The paper li...

Using Deep and Megatron to Train Megatron-Turing NLG 530B, A Large-Scale Generative Language Model

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper details the training of Megatron-Turing NLG 530B, a large-scale generative language model. It likely discusses the architectural details of...

Training Compute-Optimal Large Language Models

<h2 class="text-2xl font-bold mb-4">Summary</h2> This DeepMind paper introduces the 'Chinchilla' model, a large language model trained based on a compute-optimal scaling law. The authors challenge th...

OPT: Open Pre-trained Transformer Language Models

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper introduces OPT (Open Pre-trained Transformer), a suite of open-source, pre-trained Transformer language models developed by Meta. The model...

Unifying Language Learning Paradigms

<h2 class="text-2xl font-bold mb-4">Summary</h2> This research paper, published by Google in May 2022, introduces a novel approach to unify language learning paradigms. While the specifics are missin...

Emergent Abilities of Large Language Models

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper from Google, published in June 2022, examines the phenomenon of emergent abilities in large language models (LLMs). It focuses on capabilit...

Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language models

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper, from Google, introduces and analyzes the BIG-bench benchmark, a large-scale, diverse suite of tasks designed to evaluate the capabilities ...

Language Models are General-Purpose Interfaces

<h2 class="text-2xl font-bold mb-4">Summary</h2> This research paper from Microsoft, published in June 2022, posits that language models (LMs) can serve as general-purpose interfaces. It likely inves...

Improving alignment of dialogue agents via targeted human judgements

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper from DeepMind, likely focusing on their Sparrow dialogue agent, explores methods for improving the alignment of dialogue agents with human ...

Scaling Instruction-Finetuned Language Models

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper investigates the impact of scaling instruction-tuned language models, focusing on the Flan-T5 and PaLM families. It explores how the perfor...

Holistic Evaluation of Language Models

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper, from Stanford University, focuses on the holistic evaluation of Language Models (LMs). It likely introduces a new methodology or framework...

BLOOM: A 176B-Parameter Open-Access Multilingual Language Model

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper introduces BLOOM, a 176-billion parameter, open-access, multilingual language model. Developed by the BigScience research workshop, BLOOM r...

Galactica: A Large Language Model for Science

<h2 class="text-2xl font-bold mb-4">Summary</h2> The research paper introduces Galactica, a large language model (LLM) specifically designed for scientific tasks. Trained on a massive dataset of scie...

OPT-IML: Scaling Language Model Instruction Meta Learning through the Lens of Generalization

<h2 class="text-2xl font-bold mb-4">Summary</h2> The research paper, "OPT-IML: Scaling Language Model Instruction Meta Learning through the Lens of Generalization", focuses on improving instruction m...

The Flan Collection: Designing Data and Methods for Effective Instruction Tuning

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper, published by Google in January 2023, introduces the Flan Collection, focusing on designing data and methods for effective instruction tuni...

LLaMA: Open and Efficient Foundation Language Models

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper introduces LLaMA, a collection of foundation language models ranging from 7B to 70B parameters. These models are trained on a vast corpus o...

Language Is Not All You Need: Aligning Perception with Language Models

<h2 class="text-2xl font-bold mb-4">Summary</h2> This research paper, originating from Microsoft and likely published around February 2023, focuses on the limitations of language models and the need ...

PaLM-E: An Embodied Multimodal Language Model

<h2 class="text-2xl font-bold mb-4">Summary</h2> PaLM-E is a multimodal language model designed for embodied agents, integrating visual and language modalities for improved performance in robotics an...

GPT-4 Technical Report

<h2 class="text-2xl font-bold mb-4">Summary</h2> The GPT-4 Technical Report, published by OpenAI in March 2023, details the architecture, capabilities, and limitations of the GPT-4 large language mod...

Visual Instruction Tuning

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper likely introduces LLaVA (Language-and-Vision Assistant), a model that leverages visual instruction tuning. Visual instruction tuning involv...

Principle-Driven Self-Alignment of Language Models from Scratch with Minimal Human Supervision

<h2 class="text-2xl font-bold mb-4">Summary</h2> This research paper, published in May 2023 by CMU and others, focuses on self-alignment of large language models (LLMs) from the ground up, utilizing ...

PaLM 2 Technical Report

<h2 class="text-2xl font-bold mb-4">Summary</h2> The PaLM 2 Technical Report details the architecture, training, and capabilities of Google's PaLM 2 large language model. It likely covers aspects suc...

RWKV: Reinventing RNNs for the Transformer Era

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper introduces RWKV, a novel architecture that seeks to combine the strengths of both Recurrent Neural Networks (RNNs) and Transformers. RWKV a...

Direct Preference Optimization: Your Language Model is Secretly a Reward Model

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper introduces Direct Preference Optimization (DPO), a novel approach for aligning language models with human preferences. DPO offers a simplif...

Tree of Thoughts: Deliberate Problem Solving with Large Language Models

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper introduces the Tree of Thoughts (ToT) framework, a novel approach to enhance the problem-solving capabilities of Large Language Models (LLM...

Llama 2: Open Foundation and Fine-Tuned Chat Models

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper details the development and release of Llama 2, a family of large language models (LLMs) by Meta. It presents models ranging from 7 to 70 b...

Mistral 7B

<h2 class="text-2xl font-bold mb-4">Summary</h2> This research paper introduces Mistral 7B, a 7-billion parameter language model. The paper likely details the model's architecture, training methodolo...

Mamba: Linear-Time Sequence Modeling with Selective State Spaces

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper introduces Mamba, a novel state space model (SSM) for sequence modeling that achieves linear-time processing of sequences. Mamba addresses ...

DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model

<h2 class="text-2xl font-bold mb-4">Summary</h2> DeepSeek-V2 presents a Mixture-of-Experts (MoE) language model designed for strong performance, economic efficiency, and operational efficiency. The p...

OLMo: Accelerating the Science of Language Models

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper introduces OLMo, a language model initiative designed to accelerate the scientific understanding of language models. It likely details the ...

The Llama 3 Herd of Models

<h2 class="text-2xl font-bold mb-4">Summary</h2> The Meta paper, "The Llama 3 Herd of Models," details the development and capabilities of the Llama 3 family of large language models (LLMs). This res...

OLMoE: Open Mixture-of-Experts Language Models

<h2 class="text-2xl font-bold mb-4">Summary</h2> This paper introduces OLMoE, an open-source Mixture-of-Experts (MoE) language model developed by Ai2. It likely describes the architecture, training m...

Qwen2.5 Technical Report

<h2 class="text-2xl font-bold mb-4">Summary</h2> The Qwen2.5 Technical Report, published by Alibaba, details the advancements and improvements made in the Qwen2.5 large language model compared to its...

DeepSeek-V3 Technical Report

<h2 class="text-2xl font-bold mb-4">Summary</h2> The DeepSeek-V3 Technical Report, published by DeepSeek in December 2024, details the advancements and technical specifications of their latest large ...

DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning

<h2 class="text-2xl font-bold mb-4">Summary</h2> The DeepSeek-R1 paper introduces a novel approach to enhance reasoning capabilities in Large Language Models (LLMs) using Reinforcement Learning (RL)....