[
  {
    "title": "ImageNet Classification with Deep Convolutional Neural Networks",
    "authors": "Krizhevsky et al.",
    "year": 2012,
    "venue": "NeurIPS",
    "why": "A deep convolutional network trained on GPUs won the ImageNet competition by a large margin over prior methods. The result showed that deep neural networks could learn useful visual features directly from large labeled datasets, and it started the shift toward deep learning across the field.",
    "links": {"paper": "https://proceedings.neurips.cc/paper/2012/hash/c399862d3b9d6b76c8436e924a68c45b-Abstract.html"}
  },
  {
    "title": "Efficient Estimation of Word Representations in Vector Space",
    "authors": "Mikolov et al.",
    "year": 2013,
    "venue": "ICLR Workshop",
    "why": "Introduced the word2vec models for learning dense word embeddings efficiently from large text corpora. The learned vectors captured syntactic and semantic relationships and became a standard input representation for later natural language processing systems.",
    "links": {"paper": "https://arxiv.org/abs/1301.3781"}
  },
  {
    "title": "Deep Residual Learning for Image Recognition",
    "authors": "He et al.",
    "year": 2015,
    "venue": "CVPR",
    "why": "Introduced residual connections that let networks be trained with far greater depth than was previously practical. This addressed the degradation problem in very deep networks and the residual block became a common building component in later architectures.",
    "links": {"paper": "https://arxiv.org/abs/1512.03385", "code": "https://github.com/KaimingHe/deep-residual-networks"}
  },
  {
    "title": "Attention Is All You Need",
    "authors": "Vaswani et al.",
    "year": 2017,
    "venue": "NeurIPS",
    "why": "Introduced the Transformer, an architecture built entirely on attention mechanisms rather than recurrence or convolution. It enabled more parallel training and became the basis for most subsequent large language models.",
    "links": {"paper": "https://arxiv.org/abs/1706.03762", "code": "https://github.com/tensorflow/tensor2tensor"}
  },
  {
    "title": "Proximal Policy Optimization Algorithms",
    "authors": "Schulman et al.",
    "year": 2017,
    "venue": "arXiv",
    "why": "Proposed PPO, a reinforcement learning method that is simpler to implement and tune than earlier policy-gradient approaches while remaining stable. It became a widely used algorithm and is a common component of reinforcement learning from human feedback pipelines.",
    "links": {"paper": "https://arxiv.org/abs/1707.06347", "code": "https://github.com/openai/baselines"}
  },
  {
    "title": "Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm",
    "authors": "Silver et al.",
    "year": 2017,
    "venue": "arXiv",
    "why": "Described AlphaZero, a single reinforcement learning method that reached strong play in chess, shogi, and Go using only self-play and the game rules. It showed that a general algorithm without human game data could match or exceed specialized programs.",
    "links": {"paper": "https://arxiv.org/abs/1712.01815"}
  },
  {
    "title": "BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding",
    "authors": "Devlin et al.",
    "year": 2018,
    "venue": "NAACL",
    "why": "Introduced a Transformer encoder pre-trained with masked language modeling to produce bidirectional text representations. Fine-tuning the pre-trained model improved results across many language understanding tasks and popularized the pre-train then fine-tune approach.",
    "links": {"paper": "https://arxiv.org/abs/1810.04805", "code": "https://github.com/google-research/bert"}
  },
  {
    "title": "Scaling Laws for Neural Language Models",
    "authors": "Kaplan et al.",
    "year": 2020,
    "venue": "arXiv",
    "why": "Measured how language model loss changes with model size, dataset size, and compute, finding smooth power-law relationships. These trends provided a basis for predicting performance and planning the training of larger models.",
    "links": {"paper": "https://arxiv.org/abs/2001.08361"}
  },
  {
    "title": "Language Models are Few-Shot Learners",
    "authors": "Brown et al.",
    "year": 2020,
    "venue": "NeurIPS",
    "why": "Presented GPT-3, a 175-billion-parameter language model, and showed it could perform many tasks from instructions and a few examples in the prompt without gradient updates. This in-context learning behavior influenced how later models were used and evaluated.",
    "links": {"paper": "https://arxiv.org/abs/2005.14165"}
  },
  {
    "title": "Chain-of-Thought Prompting Elicits Reasoning in Large Language Models",
    "authors": "Wei et al.",
    "year": 2022,
    "venue": "NeurIPS",
    "why": "Showed that prompting a model to produce intermediate reasoning steps improves accuracy on arithmetic, commonsense, and symbolic tasks. The effect appeared mainly in sufficiently large models and became a common prompting technique.",
    "links": {"paper": "https://arxiv.org/abs/2201.11903"}
  },
  {
    "title": "Training language models to follow instructions with human feedback",
    "authors": "Ouyang et al.",
    "year": 2022,
    "venue": "NeurIPS",
    "why": "Described InstructGPT, which fine-tuned a language model using supervised examples and reinforcement learning from human feedback to better follow user instructions. Human raters preferred its outputs over the much larger base model, and the approach shaped later assistant-style systems.",
    "links": {"paper": "https://arxiv.org/abs/2203.02155"}
  },
  {
    "title": "Training Compute-Optimal Large Language Models",
    "authors": "Hoffmann et al.",
    "year": 2022,
    "venue": "NeurIPS",
    "why": "Found that many large models were undertrained and that for a fixed compute budget model size and training tokens should scale in roughly equal proportion. The Chinchilla model trained under this guidance outperformed larger models, changing how training budgets were allocated.",
    "links": {"paper": "https://arxiv.org/abs/2203.15556"}
  },
  {
    "title": "ReAct: Synergizing Reasoning and Acting in Language Models",
    "authors": "Yao et al.",
    "year": 2022,
    "venue": "ICLR",
    "why": "Combined step-by-step reasoning with actions such as querying external tools within a single prompting loop. This interleaving reduced fabricated answers on knowledge tasks and became a common pattern for building language model agents.",
    "links": {"paper": "https://arxiv.org/abs/2210.03629", "code": "https://github.com/ysymyth/ReAct"}
  },
  {
    "title": "Constitutional AI: Harmlessness from AI Feedback",
    "authors": "Bai et al.",
    "year": 2022,
    "venue": "arXiv",
    "why": "Proposed training a model to be less harmful using a written set of principles and AI-generated feedback, reducing reliance on human labels for harmful content. The method uses model self-critique and revision to guide behavior toward the stated principles.",
    "links": {"paper": "https://arxiv.org/abs/2212.08073"}
  },
  {
    "title": "DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning",
    "authors": "DeepSeek-AI",
    "year": 2025,
    "venue": "Nature",
    "why": "Showed that reasoning behavior such as self-checking and longer solution traces could be developed largely through reinforcement learning on verifiable tasks rather than supervised reasoning examples. The authors released the models and distilled smaller versions under an open license.",
    "links": {"paper": "https://arxiv.org/abs/2501.12948", "code": "https://github.com/deepseek-ai/DeepSeek-R1"}
  }
]