papers.sample.json

{
  "papers": [
    {
      "id": "paper-001",
      "rank": 1,
      "alias": "AlphaNet",
      "title": "AlphaNet: Neural Architecture for Distributed Learning",
      "url": "https://example.com/paper-001",
      "year": 2023,
      "score": 98,
      "categories": ["Architecture", "Distributed"],
      "cluster": "infra",
      "reason": "A foundational architecture for distributed neural networks",
      "has_review": true,
      "one_liner": "Distributed learning made simple",
      "key_discoveries": ["Async gradient sync", "Linear scaling efficiency"],
      "review_url": "https://example.com/review/paper-001",
      "reason_en": "A foundational architecture for distributed neural networks"
    },
    {
      "id": "paper-002",
      "rank": 2,
      "alias": "BetaLM",
      "title": "BetaLM: Efficient Language Modeling at Scale",
      "url": "https://example.com/paper-002",
      "year": 2023,
      "score": 96,
      "categories": ["LLM", "Scaling"],
      "cluster": "llm",
      "reason": "Efficient scaling laws for language models",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "Efficient scaling laws for language models"
    },
    {
      "id": "paper-003",
      "rank": 3,
      "alias": "GammaVision",
      "title": "GammaVision: Self-Supervised Visual Representation Learning",
      "url": "https://example.com/paper-003",
      "year": 2022,
      "score": 95,
      "categories": ["Vision", "SelfSupervised"],
      "cluster": "vision",
      "reason": "Label-free visual features that rival supervised methods",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "Label-free visual features that rival supervised methods"
    },
    {
      "id": "paper-004",
      "rank": 4,
      "alias": "DeltaGen",
      "title": "DeltaGen: Controllable Image Generation with Diffusion",
      "url": "https://example.com/paper-004",
      "year": 2023,
      "score": 94,
      "categories": ["Diffusion", "Generation"],
      "cluster": "generation",
      "reason": "Fine-grained control over diffusion image synthesis",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "Fine-grained control over diffusion image synthesis"
    },
    {
      "id": "paper-005",
      "rank": 5,
      "alias": "EpsilonAlign",
      "title": "EpsilonAlign: Reward Modeling for Safer AI",
      "url": "https://example.com/paper-005",
      "year": 2024,
      "score": 93,
      "categories": ["RLHF", "Safety"],
      "cluster": "alignment",
      "reason": "Improved reward modeling for AI alignment",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "Improved reward modeling for AI alignment"
    },
    {
      "id": "paper-006",
      "rank": 6,
      "alias": "ZetaFusion",
      "title": "ZetaFusion: Multimodal Understanding Through Cross-Attention",
      "url": "https://example.com/paper-006",
      "year": 2024,
      "score": 93,
      "categories": ["Multimodal", "Vision", "NLP"],
      "cluster": "multimodal",
      "reason": "Unified cross-modal attention for text, image, and audio",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "Unified cross-modal attention for text, image, and audio"
    },
    {
      "id": "paper-007",
      "rank": 7,
      "alias": "EtaCode",
      "title": "EtaCode: LLM-Powered Code Generation Agent",
      "url": "https://example.com/paper-007",
      "year": 2024,
      "score": 92,
      "categories": ["CodeGen", "Agent"],
      "cluster": "agent",
      "reason": "Autonomous coding agent with tool-use capabilities",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "Autonomous coding agent with tool-use capabilities"
    },
    {
      "id": "paper-008",
      "rank": 8,
      "alias": "ThetaFlash",
      "title": "ThetaFlash: Hardware-Aware Attention Optimization",
      "url": "https://example.com/paper-008",
      "year": 2023,
      "score": 92,
      "categories": ["Attention", "Efficient"],
      "cluster": "efficient",
      "reason": "IO-aware attention reducing memory reads by 10x",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "IO-aware attention reducing memory reads by 10x"
    },
    {
      "id": "paper-009",
      "rank": 9,
      "alias": "IotaReason",
      "title": "IotaReason: Chain-of-Thought Reasoning in LLMs",
      "url": "https://example.com/paper-009",
      "year": 2023,
      "score": 91,
      "categories": ["LLM", "Reasoning"],
      "cluster": "llm",
      "reason": "Systematic chain-of-thought improves complex reasoning by 40%",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "Systematic chain-of-thought improves complex reasoning by 40%"
    },
    {
      "id": "paper-010",
      "rank": 10,
      "alias": "KappaDetect",
      "title": "KappaDetect: Real-Time Object Detection with Transformers",
      "url": "https://example.com/paper-010",
      "year": 2022,
      "score": 91,
      "categories": ["ObjectDetection", "Vision"],
      "cluster": "vision",
      "reason": "End-to-end transformer detection at 60 FPS",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "End-to-end transformer detection at 60 FPS"
    },
    {
      "id": "paper-011",
      "rank": 11,
      "alias": "LambdaLoRA",
      "title": "LambdaLoRA: Parameter-Efficient Fine-Tuning",
      "url": "https://example.com/paper-011",
      "year": 2023,
      "score": 90,
      "categories": ["PEFT", "LLM"],
      "cluster": "efficient",
      "reason": "Fine-tune billion-parameter models with 0.1% trainable params",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "Fine-tune billion-parameter models with 0.1% trainable params"
    },
    {
      "id": "paper-012",
      "rank": 12,
      "alias": "MuDiffuse",
      "title": "MuDiffuse: Music Generation via Latent Diffusion",
      "url": "https://example.com/paper-012",
      "year": 2024,
      "score": 90,
      "categories": ["Audio", "Diffusion"],
      "cluster": "generation",
      "reason": "Text-to-music with latent diffusion and melody control",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "Text-to-music with latent diffusion and melody control"
    },
    {
      "id": "paper-013",
      "rank": 13,
      "alias": "NuAgent",
      "title": "NuAgent: Tool-Augmented Autonomous Agents",
      "url": "https://example.com/paper-013",
      "year": 2024,
      "score": 89,
      "categories": ["Agent", "ToolUse"],
      "cluster": "agent",
      "reason": "Agents that plan, search, and execute multi-step tasks",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "Agents that plan, search, and execute multi-step tasks"
    },
    {
      "id": "paper-014",
      "rank": 14,
      "alias": "OmegaNorm",
      "title": "OmegaNorm: Rethinking Normalization for Deep Networks",
      "url": "https://example.com/paper-014",
      "year": 2022,
      "score": 89,
      "categories": ["Normalization", "Training"],
      "cluster": "infra",
      "reason": "Unified normalization layer that adapts per-layer statistics",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "Unified normalization layer that adapts per-layer statistics"
    },
    {
      "id": "paper-015",
      "rank": 15,
      "alias": "PiGuard",
      "title": "PiGuard: Constitutional AI for Content Safety",
      "url": "https://example.com/paper-015",
      "year": 2024,
      "score": 88,
      "categories": ["Safety", "RLHF"],
      "cluster": "alignment",
      "reason": "Self-supervised safety classifier trained via constitution",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "Self-supervised safety classifier trained via constitution"
    },
    {
      "id": "paper-016",
      "rank": 16,
      "alias": "RhoEmbed",
      "title": "RhoEmbed: Universal Text Embeddings",
      "url": "https://example.com/paper-016",
      "year": 2023,
      "score": 88,
      "categories": ["Embeddings", "NLP"],
      "cluster": "llm",
      "reason": "One embedding model for retrieval, classification, and clustering",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "One embedding model for retrieval, classification, and clustering"
    },
    {
      "id": "paper-017",
      "rank": 17,
      "alias": "SigmaSeg",
      "title": "SigmaSeg: Open-Vocabulary Segmentation",
      "url": "https://example.com/paper-017",
      "year": 2023,
      "score": 87,
      "categories": ["Segmentation", "Vision"],
      "cluster": "vision",
      "reason": "Segment anything with free-form text prompts",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "Segment anything with free-form text prompts"
    },
    {
      "id": "paper-018",
      "rank": 18,
      "alias": "TauVideo",
      "title": "TauVideo: Long-Form Video Understanding with LLMs",
      "url": "https://example.com/paper-018",
      "year": 2024,
      "score": 87,
      "categories": ["Video", "Multimodal"],
      "cluster": "multimodal",
      "reason": "Hour-long video QA through temporal token compression",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "Hour-long video QA through temporal token compression"
    },
    {
      "id": "paper-019",
      "rank": 19,
      "alias": "UpsilonQuant",
      "title": "UpsilonQuant: 4-bit Quantization Without Quality Loss",
      "url": "https://example.com/paper-019",
      "year": 2023,
      "score": 86,
      "categories": ["Quantization", "Efficient"],
      "cluster": "efficient",
      "reason": "Activation-aware weight quantization preserving 99% accuracy",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "Activation-aware weight quantization preserving 99% accuracy"
    },
    {
      "id": "paper-020",
      "rank": 20,
      "alias": "PhiRobot",
      "title": "PhiRobot: Vision-Language-Action Model for Robotics",
      "url": "https://example.com/paper-020",
      "year": 2024,
      "score": 86,
      "categories": ["Robotics", "Multimodal"],
      "cluster": "agent",
      "reason": "Unified VLA model bridging language instructions to robot actions",
      "has_review": false,
      "one_liner": "",
      "key_discoveries": [],
      "review_url": "",
      "reason_en": "Unified VLA model bridging language instructions to robot actions"
    }
  ],
  "edges": [
    {"source": "paper-001", "target": "paper-002", "type": "lineage"},
    {"source": "paper-001", "target": "paper-003", "type": "lineage"},
    {"source": "paper-001", "target": "paper-008", "type": "lineage"},
    {"source": "paper-002", "target": "paper-009", "type": "lineage"},
    {"source": "paper-002", "target": "paper-011", "type": "lineage"},
    {"source": "paper-002", "target": "paper-016", "type": "lineage"},
    {"source": "paper-003", "target": "paper-010", "type": "lineage"},
    {"source": "paper-003", "target": "paper-017", "type": "lineage"},
    {"source": "paper-004", "target": "paper-012", "type": "lineage"},
    {"source": "paper-005", "target": "paper-015", "type": "lineage"},
    {"source": "paper-006", "target": "paper-018", "type": "lineage"},
    {"source": "paper-006", "target": "paper-020", "type": "lineage"},
    {"source": "paper-007", "target": "paper-013", "type": "lineage"},
    {"source": "paper-008", "target": "paper-019", "type": "lineage"},
    {"source": "paper-001", "target": "paper-006", "type": "lineage"},
    {"source": "paper-001", "target": "paper-007", "type": "lineage"},
    {"source": "paper-001", "target": "paper-014", "type": "lineage"},
    {"source": "paper-005", "target": "paper-007", "type": "lineage"},
    {"source": "paper-003", "target": "paper-006", "type": "lineage"}
  ],
  "clusters": {
    "llm": {"label": "LLM", "color": "#3b82f6", "icon": "\ud83e\udde0"},
    "infra": {"label": "Architecture", "color": "#8b5cf6", "icon": "\ud83c\udfd7\ufe0f"},
    "vision": {"label": "Vision", "color": "#10b981", "icon": "\ud83d\udc41\ufe0f"},
    "generation": {"label": "Generation", "color": "#f59e0b", "icon": "\ud83c\udfa8"},
    "multimodal": {"label": "Multimodal", "color": "#ec4899", "icon": "\ud83d\udd17"},
    "alignment": {"label": "Alignment", "color": "#ef4444", "icon": "\ud83d\udee1\ufe0f"},
    "agent": {"label": "Agent & Code", "color": "#f97316", "icon": "\ud83e\udd16"},
    "efficient": {"label": "Efficient", "color": "#14b8a6", "icon": "\u26a1"}
  }
}