{
  "docId": "019dd923-5e88-73ef-bd5d-06b04d219fea",
  "docSlug": "dd91c78f6570bf29",
  "documentTitle": "2023 Air Street Capital The State of AI Report 2023",
  "authorId": "AirStreetCapital",
  "authorName": "Air Street Capital",
  "documentKindSlug": "consulting-deck",
  "documentKindLabel": "Consulting deck",
  "sourceTypeSlug": "vc_research",
  "sourceTypeLabel": "VC research",
  "presentationDate": null,
  "orientation": "landscape",
  "aspectRatio": 1.777,
  "pageNumber": 25,
  "pageCount": 163,
  "prevPage": 24,
  "nextPage": 26,
  "slideType": "industry_trends",
  "function": "summarize",
  "density": "dense",
  "nDataPoints": 3,
  "notes": "Part of the State of AI 2023 report.",
  "elementsJson": [
    "headline_text",
    "bullet_list",
    "paragraph"
  ],
  "metadataConfidence": 1,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5e88-73ef-bd5d-06b04d219fea/25",
  "deckHref": "/decks/019dd923-5e88-73ef-bd5d-06b04d219fea",
  "deckJsonHref": "/decks/019dd923-5e88-73ef-bd5d-06b04d219fea.json",
  "deckAnchorHref": "/decks/019dd923-5e88-73ef-bd5d-06b04d219fea#slide-25",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "Increased context length and large datasets require architectural innovations.",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd952-47c4-7719-a222-6bd73f292178",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.6,
        "w": 0.96,
        "x": 0.02,
        "y": 0.28
      },
      "kind": "list",
      "text": "FlashAttention introduces a significant memory saving by making attention linear instead of quadratic in sequence length. FlashAttention-2 further improves computing the attention matrix by having fewer non-matmul FLOPs, better parallelism and better work partitioning. The result is a 2.8x training speedup of GPT-style models.\nReducing the number of bits in the parameters reduces both the memory footprint and the latency of LLMs. The case for 4-bit precision: k-bit Inference Scaling Laws shows across a variety of LLMs that 4-bit quantisation is universally optimal for maximizing zero-shot accuracy and reducing the number of bits used.\nSpeculative decoding enables decoding multiple tokens in parallel through multiple model heads rather than forward passes, speeding up inference by 2-3X for certain models.\nSWARM Parallelism is a training algorithm designed for poorly connected and unreliable devices. It enables training billion-scale LLMs on low bandwidth networks and low-power GPUs while achieving high training throughput.",
      "attrs": null,
      "subkind": "bullet",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "0bbf9d15-c8c5-48c6-a409-9cae0c875f2f",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": null,
      "kind": "metric",
      "text": "training speedup: 2.8x",
      "attrs": null,
      "subkind": "primary",
      "toolName": "Quantification",
      "toolSlug": "quantification",
      "confidence": null,
      "componentId": "019dd952-47c4-7719-a222-6f86d8a65f8b",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.03,
        "w": 0.65,
        "x": 0.02,
        "y": 0.22
      },
      "kind": "paragraph",
      "text": "Increased context length and large datasets require architectural innovations.",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "7bbf3d2c-3416-4cfd-a70a-eb733d78f305",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.04,
        "w": 0.4,
        "x": 0.02,
        "y": 0.14
      },
      "kind": "title",
      "text": "Keeping up with high memory demands",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "ed185c09-dc09-4369-840f-4c7b92cd9bdd",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [
    {
      "name": "List presentation",
      "slug": "list-presentation",
      "agent": null,
      "layer": "slide",
      "matchId": "1b39d50a-5ffe-4405-aca5-dd52669e3f8c",
      "evidence": "The slide contains a list of bullet points discussing architectural innovations for increased context length and large datasets.",
      "confidence": 0.7
    }
  ],
  "frameworks": [],
  "arcBeats": [
    {
      "to": 120,
      "from": 11,
      "beatId": "019dd95a-0682-776c-8e35-41afd44ef59f",
      "arcName": "The Triple Take",
      "arcSlug": "triple-take",
      "beatName": "The Facts (What)",
      "beatSlug": "triple-take-the-facts-what",
      "evidence": "Research + Industry sections inventory model, compute, funding facts.",
      "position": 1,
      "confidence": 78,
      "parentBeatName": "Setup",
      "parentBeatSlug": "setup"
    },
    {
      "to": 120,
      "from": 11,
      "beatId": "019dd95a-0682-776c-8e35-523bfb7f96e6",
      "arcName": "The Mountain",
      "arcSlug": "mountain",
      "beatName": "Rising Action",
      "beatSlug": "mountain-rising-action",
      "evidence": "Escalating capabilities, compute concentration and capital flows.",
      "position": 2,
      "confidence": 45,
      "parentBeatName": "Development",
      "parentBeatSlug": "development"
    }
  ],
  "loops": [
    {
      "to": 31,
      "from": 22,
      "name": "Zoom In",
      "slug": "06-zoom-in",
      "bestFor": "Technical deep-dives, case studies, detailed analysis",
      "matchId": "019dd95a-07fe-70ce-8d3d-ea04ad01cb62",
      "evidence": "Slides drill from emergent-capability mirage into context-length, memory and data-exhaustion details.",
      "position": 3,
      "objective": "Zoom into the data and context-length bottleneck",
      "structure": "The Big Picture -> Key Area of Focus -> Specific Detail -> Implication",
      "confidence": 72,
      "description": "Start broad, then progressively focus on specific details that prove your point"
    }
  ],
  "imagePathAlt": null,
  "thumbSrc": null,
  "thumbSrcAlt": null,
  "locked": true
}