The paper “Attention Is All You Need” (2017) introduced the Transformer — a novel neural architecture relying solely on self-attention, removing recurrence and convolutions. It revolutionized machine translation by dramatically improving training speed and translation quality (e.g., achieving 28.4 BLEU on English-German tasks), setting new state-of-the-art benchmarks. Its modular, parallelizable design opened the door to large-scale pretraining and fine-tuning, ultimately laying the foundation for modern large language models like BERT and GPT. This paper reshaped the landscape of NLP and deep learning, making attention-based models the dominant paradigm across many tasks.
This paper introduces GPT-2, showing that large-scale language models trained on diverse internet text can perform a wide range of natural language tasks in a zero-shot setting — without any task-specific training. By scaling up to 1.5 billion parameters and training on WebText, GPT-2 achieves state-of-the-art or competitive results on benchmarks like language modeling, reading comprehension, and question answering. Its impact has been profound, pioneering the trend toward general-purpose, unsupervised language models and paving the way for today’s foundation models in AI.
reveals that language model performance improves predictably as you scale up model size, dataset size, and compute, following smooth power-law relationships. It shows that larger models are more sample-efficient, and optimally efficient training uses very large models on moderate data, stopping well before convergence. The work provided foundational insights that influenced the development of massive models like GPT-3 and beyond, shaping how the AI community understands trade-offs between size, data, and compute in building ever-stronger models.
This paper introduces GPT-3, a 175-billion-parameter autoregressive language model that achieves impressive zero-shot, one-shot, and few-shot performance across diverse NLP tasks without task-specific fine-tuning. Its scale allows it to generalize from natural language prompts, rivaling or surpassing prior state-of-the-art models that require fine-tuning. The paper’s impact is profound: it demonstrated the power of scaling laws, reshaped research on few-shot learning, and sparked widespread adoption of large-scale language models, influencing advancements in AI applications, ethical debates, and commercial deployments globally.
This paper introduces GPT-4, a large multimodal model that processes both text and images, achieving human-level performance on many academic and professional benchmarks like the bar exam and GRE. It significantly advances language understanding, multilingual capabilities, and safety alignment over previous models, outperforming GPT-3.5 by wide margins. Its impact is profound, setting new standards for natural language processing, enabling safer and more powerful applications, and driving critical research on scaling laws, safety, bias, and the societal implications of AI deployment.
This paper presents DeepSeek-V2, a 236B-parameter open-source Mixture-of-Experts (MoE) language model that activates only 21B parameters per token, achieving top-tier bilingual (English and Chinese) performance with remarkable training cost savings (42.5%) and inference efficiency (5.76× throughput) compared to previous models. Its innovations—Multi-head Latent Attention (MLA) and DeepSeekMoE—reduce memory bottlenecks and boost specialization. The paper’s impact lies in advancing economical, efficient large-scale language modeling, pushing open-source models closer to closed-source leaders, and paving the way for future multimodal and AGI-aligned systems.
This paper introduces DeepSeek-V3, a 671B-parameter Mixture-of-Experts (MoE) language model that activates only 37B parameters per token for efficient training and inference. By leveraging innovations like Multi-head Latent Attention, auxiliary-loss-free load balancing, and multi-token prediction, it achieves top-tier performance across math, code, multilingual, and reasoning tasks. Despite its massive scale, DeepSeek-V3 maintains economical training costs and outperforms all other open-source models, achieving results comparable to leading closed-source models like GPT-4o and Claude-3.5, thereby significantly narrowing the open-source vs. closed-source performance gap.
This paper introduces DeepSeek-R1, a large language model that improves reasoning purely through reinforcement learning (RL), even without supervised fine-tuning. It shows that reasoning skills like chain-of-thought, self-reflection, and verification can naturally emerge from RL, achieving performance comparable to OpenAI’s top models. Its distilled smaller models outperform many open-source alternatives, democratizing advanced reasoning for smaller systems. The work impacts the field by proving RL-alone reasoning is viable and by open-sourcing both large and distilled models, opening new directions for scalable, cost-effective LLM training and future development in reasoning-focused AI systems.
