{
  "schema_version": "1.1",
  "id": "archive:https://www.youtube.com/watch?v=WiqDvX6isc4",
  "slug": "scaling-compute-on-context-jack-morris-engram-1pxr9bt",
  "url": "https://feed7.dev/p/scaling-compute-on-context-jack-morris-engram-1pxr9bt",
  "title": "Scaling Compute on Context — Jack Morris, Engram",
  "why_included": "Training directly on a private corpus can memorize it without producing useful behavior. Durable context learning likely needs synthetic study tasks and recursive self-improvement, not a single fine-tuning pass.",
  "summary": "Morris frames the problem as teaching a pretrained model an unstructured private corpus. Direct next-token training on **10,000 financial reports** can drive loss to **0.00001** yet still produce collapsed generations, showing that memorization is not usable knowledge.",
  "practical_implication": "For private code, messages, or documents, distinguish retrieval from learning. Context compression, on-policy distillation, synthetic question generation, and continued pretraining offer different tradeoffs; the desired system should keep generating harder study material as the model improves.",
  "agent_context": "Morris frames the problem as teaching a pretrained model an unstructured private corpus. Direct next-token training on **10,000 financial reports** can drive loss to **0.00001** yet still produce collapsed generations, showing that memorization is not usable knowledge.\n\nFor private code, messages, or documents, distinguish retrieval from learning. Context compression, on-policy distillation, synthetic question generation, and continued pretraining offer different tradeoffs; the desired system should keep generating harder study material as the model improves.\n\nEvery described method has limits: context methods require the data to fit, synthetic training can overwrite prior knowledge, and one-pass approaches plateau. Recursive self-improvement is presented as the research target, not a settled recipe.",
  "source": {
    "name": "AI Engineer",
    "url": "https://www.youtube.com/watch?v=WiqDvX6isc4",
    "published_at": "2026-08-12T15:30:14.000Z"
  },
  "source_class": "video",
  "content_type": "Video",
  "layer": "context",
  "domains": [
    "data",
    "research"
  ],
  "topics": [
    "context-engineering",
    "retrieval",
    "agent-memory"
  ],
  "verification": {
    "status": "source_linked",
    "label": "Source Linked",
    "method": "source_feed",
    "verified_at": null
  },
  "uncertainty": [
    "Every described method has limits: context methods require the data to fit, synthetic training can overwrite prior knowledge, and one-pass approaches plateau. Recursive self-improvement is presented as the research target, not a settled recipe."
  ],
  "connected_context": {
    "meaning": "This rules out low training loss as evidence that a private corpus has become usable knowledge and turns corpus adaptation into a choice among retrieval, compression, and parameter updates. It reinforces hybrid memory designs and workload-specific evaluation, while warning that consolidation can overwrite prior knowledge and that recursive improvement remains an open target.",
    "corpus_size": 468,
    "generated_at": "2026-08-17T10:04:35.619Z",
    "connections": [
      {
        "title": "UniMem: Complementary Episodic-to-Parametric Memory for Boundary-Agnostic Task Streams",
        "source_name": "arXiv",
        "source_url": "https://arxiv.org/abs/2607.26017v1",
        "feed7_url": "https://feed7.dev/p/2607-26017v1-1opv1da",
        "reason": "UniMem operationalizes the retrieval-versus-learning distinction by routing novel episodes to retrieval and recurring patterns to parameters, but does not settle overwrite risk."
      },
      {
        "title": "Teaching Nemotron Greek: Mining a Corpus, Adapting Retrieval, and Grounding Generation for Modern Greek across Specialist Domains",
        "source_name": "arXiv",
        "source_url": "https://arxiv.org/abs/2608.05138v1",
        "feed7_url": "https://feed7.dev/p/2608-05138v1-0bvu6le",
        "reason": "The Greek RAG study reinforces the need to evaluate corpus access methods on the actual domain rather than assume learned or dense representations will outperform lexical retrieval."
      },
      {
        "title": "Structured Memory for Edge Language Models: Persistent Context and Corpus Retrieval via O(1) SSM State Injection",
        "source_name": "arXiv",
        "source_url": "https://arxiv.org/abs/2608.02560v1",
        "feed7_url": "https://feed7.dev/p/2608-02560v1-1u8ls14",
        "reason": "PRECOG offers an architecture-specific way to reuse corpus context without retraining or repeated ingestion, covering the retrieval side of the tradeoff."
      },
      {
        "title": "Citation Needed: Provenance for LLM-Built Knowledge Graphs — Daniel Chalef, Zep AI",
        "source_name": "AI Engineer",
        "source_url": "https://www.youtube.com/watch?v=H7puB0RwJMM",
        "feed7_url": "https://feed7.dev/p/citation-needed-provenance-for-llm-built-knowledge-graphs-daniel-chalef-1iob5t8",
        "reason": "Provenance becomes an implementation requirement when compressed, synthesized, or learned corpus knowledge can no longer be traced through simple source pointers."
      }
    ]
  },
  "lifecycle": "Current",
  "published_at": "2026-08-12T15:30:14.000Z",
  "modified_at": "2026-08-12T15:30:14.000Z",
  "supersedes": [],
  "expires_at": null,
  "formats": {
    "html": "https://feed7.dev/p/scaling-compute-on-context-jack-morris-engram-1pxr9bt",
    "json": "https://feed7.dev/p/scaling-compute-on-context-jack-morris-engram-1pxr9bt.json",
    "markdown": "https://feed7.dev/p/scaling-compute-on-context-jack-morris-engram-1pxr9bt.md"
  }
}