{ "method": "Actual CUA interface captures, edited into a sequence with reading pauses and an original instrumental score. Not an uninterrupted recording or a continuous shared artifact across all specimens. Each frame retains its observed build identity.", "durationS": 60, "resolution": [ 1920, 1080 ], "fps": 30, "captureBuild": { "schemaVersion": 1, "edition": "1.2.0", "commit": "b3c87e188a41f68def59031e39c8f89b30f4ad95", "dirty": true, "sourceSha256": "347aadc4a9ef25b674c300c5ffede5a79147692e360a5aaad5411894049dd67a", "contentSha256": "88dcced3116ee6cf691d8d944adb98c78ffc108eb87720bd8224b600e8934749", "modelVersions": { "investigations": "atlas-investigations-1", "processes": "process-mechanisms-1", "learningPrompts": "investigation-prompts-1", "architecture": "architecture-execution-1", "fabrication": "atlas-fabrication-flow-1", "operatingHall": "1", "realization": "atlas-realization-1", "memory": "atlas-memory-interfaces-1", "sensor": "sensor-logistic-1", "retrieval": "lexical-cited-answer-1", "orbital": "1", "progress": "progress-evidence-2", "studioRecords": "atlas-studio-record-1", "memoryWorkload": "memory-workload-bridge-1", "factory": "factory-actual-checkpoint-1", "dataLineage": "data-lineage-1", "evaluation": "identified-evaluation-1", "applicationDecisions": "application-decisions-1", "processCell": "process-cell-acceptance-1", "speculativeVerification": "authored-draft-target-greedy-1", "serving": "causal-kv-serving-1", "learnedRetrieval": "atlas-incident-learned-1", "distributedLearning": "distributed-causal-gradient-2", "aiFamilies": "ai-families-1", "commissioning": "commissioning-execution-1", "control": "closed-loop-1", "hardwareReliability": "hardware-reliability-1", "foundations": "foundations-1", "modelLifecycle": "causal-decoder-lifecycle-1", "automation": "atlas-automation-1", "machineExecution": "machine-execution-1" }, "builtAt": "2026-09-07T21:51:56.182Z" }, "captureBuildScope": "Initial capture build shown here; individual frames retain their exact build, including the refreshed factory frame. Later atlas builds do not relabel this evidence.", "frames": [ { "file": "00-cell.png", "capturedAt": "2026-09-07T21:57:21.258Z", "sha256": "96903f52bde9316d2089bd15a5eb56a9a4dfcac641b7454c3ccde547c4741f2b", "captureBuild": { "schemaVersion": 1, "edition": "1.2.0", "commit": "b3c87e188a41f68def59031e39c8f89b30f4ad95", "dirty": true, "sourceSha256": "347aadc4a9ef25b674c300c5ffede5a79147692e360a5aaad5411894049dd67a", "contentSha256": "88dcced3116ee6cf691d8d944adb98c78ffc108eb87720bd8224b600e8934749", "modelVersions": { "investigations": "atlas-investigations-1", "processes": "process-mechanisms-1", "learningPrompts": "investigation-prompts-1", "architecture": "architecture-execution-1", "fabrication": "atlas-fabrication-flow-1", "operatingHall": "1", "realization": "atlas-realization-1", "memory": "atlas-memory-interfaces-1", "sensor": "sensor-logistic-1", "retrieval": "lexical-cited-answer-1", "orbital": "1", "progress": "progress-evidence-2", "studioRecords": "atlas-studio-record-1", "memoryWorkload": "memory-workload-bridge-1", "factory": "factory-actual-checkpoint-1", "dataLineage": "data-lineage-1", "evaluation": "identified-evaluation-1", "applicationDecisions": "application-decisions-1", "processCell": "process-cell-acceptance-1", "speculativeVerification": "authored-draft-target-greedy-1", "serving": "causal-kv-serving-1", "learnedRetrieval": "atlas-incident-learned-1", "distributedLearning": "distributed-causal-gradient-2", "aiFamilies": "ai-families-1", "commissioning": "commissioning-execution-1", "control": "closed-loop-1", "hardwareReliability": "hardware-reliability-1", "foundations": "foundations-1", "modelLifecycle": "causal-decoder-lifecycle-1", "automation": "atlas-automation-1", "machineExecution": "machine-execution-1" }, "builtAt": "2026-09-07T21:51:56.182Z" }, "lesson": "oxide-to-cell-acceptance", "studio": "cell", "observation": "INV-01 nominal10nm oxide with four following inverter loads; evaluated230.996fF output load, DC unchanged,2.4774ns loaded timing exceeds1ns; analytic synthetic2um cell.", "start": 0, "end": 6, "duration": 6, "chapter": "01 / MATTER", "title": "Geometry becomes behavior", "caption": "Change a gate load. Follow the same oxide into capacitance, logic margins and loaded delay.", "transcript": "INV-01 keeps its 10 nm oxide while four following inverter loads raise the output capacitance. Its static logic still passes, but loaded delay exceeds the chosen timing limit. This is an analytic long-channel specimen, not foundry qualification." }, { "file": "01-memory.png", "capturedAt": "2026-09-07T21:59:09.348Z", "sha256": "66ad80ce7a5fd77d37a67a0e892df84c223c01dfd7b5b2d10adf9c4a86dca13d", "captureBuild": { "schemaVersion": 1, "edition": "1.2.0", "commit": "b3c87e188a41f68def59031e39c8f89b30f4ad95", "dirty": true, "sourceSha256": "347aadc4a9ef25b674c300c5ffede5a79147692e360a5aaad5411894049dd67a", "contentSha256": "88dcced3116ee6cf691d8d944adb98c78ffc108eb87720bd8224b600e8934749", "modelVersions": { "investigations": "atlas-investigations-1", "processes": "process-mechanisms-1", "learningPrompts": "investigation-prompts-1", "architecture": "architecture-execution-1", "fabrication": "atlas-fabrication-flow-1", "operatingHall": "1", "realization": "atlas-realization-1", "memory": "atlas-memory-interfaces-1", "sensor": "sensor-logistic-1", "retrieval": "lexical-cited-answer-1", "orbital": "1", "progress": "progress-evidence-2", "studioRecords": "atlas-studio-record-1", "memoryWorkload": "memory-workload-bridge-1", "factory": "factory-actual-checkpoint-1", "dataLineage": "data-lineage-1", "evaluation": "identified-evaluation-1", "applicationDecisions": "application-decisions-1", "processCell": "process-cell-acceptance-1", "speculativeVerification": "authored-draft-target-greedy-1", "serving": "causal-kv-serving-1", "learnedRetrieval": "atlas-incident-learned-1", "distributedLearning": "distributed-causal-gradient-2", "aiFamilies": "ai-families-1", "commissioning": "commissioning-execution-1", "control": "closed-loop-1", "hardwareReliability": "hardware-reliability-1", "foundations": "foundations-1", "modelLifecycle": "causal-decoder-lifecycle-1", "automation": "atlas-automation-1", "machineExecution": "machine-execution-1" }, "builtAt": "2026-09-07T21:51:56.182Z" }, "lesson": "hbm-channel-generation-ras", "studio": "memory-workload", "observation": "Selected B[0,0] bytes01 00 00 00; two flipped codeword data bits yield syndrome6/overall parity0; SECDED blocks at40ns without published C, though stack admission passes.", "start": 6, "end": 12, "duration": 6, "chapter": "02 / MEMORY", "title": "A read must earn trust", "caption": "Follow actual operand bytes. A detected double-bit error prevents publication of the matrix result.", "transcript": "The selected B word moves through declared address translation and a real teaching DRAM trace. A separately injected double-bit codeword fault is detected at 40 modeled nanoseconds, withholding output. Unchecked consumption instead produces an independently detected wrong matrix. Stack qualification is a separate gate." }, { "file": "02-data.png", "capturedAt": "2026-09-07T21:54:53.290Z", "sha256": "e753eb169164a37af20336dfd60a925bd728a96e1b4bf59c2a274977f18d1743", "captureBuild": { "schemaVersion": 1, "edition": "1.2.0", "commit": "b3c87e188a41f68def59031e39c8f89b30f4ad95", "dirty": true, "sourceSha256": "347aadc4a9ef25b674c300c5ffede5a79147692e360a5aaad5411894049dd67a", "contentSha256": "88dcced3116ee6cf691d8d944adb98c78ffc108eb87720bd8224b600e8934749", "modelVersions": { "investigations": "atlas-investigations-1", "processes": "process-mechanisms-1", "learningPrompts": "investigation-prompts-1", "architecture": "architecture-execution-1", "fabrication": "atlas-fabrication-flow-1", "operatingHall": "1", "realization": "atlas-realization-1", "memory": "atlas-memory-interfaces-1", "sensor": "sensor-logistic-1", "retrieval": "lexical-cited-answer-1", "orbital": "1", "progress": "progress-evidence-2", "studioRecords": "atlas-studio-record-1", "memoryWorkload": "memory-workload-bridge-1", "factory": "factory-actual-checkpoint-1", "dataLineage": "data-lineage-1", "evaluation": "identified-evaluation-1", "applicationDecisions": "application-decisions-1", "processCell": "process-cell-acceptance-1", "speculativeVerification": "authored-draft-target-greedy-1", "serving": "causal-kv-serving-1", "learnedRetrieval": "atlas-incident-learned-1", "distributedLearning": "distributed-causal-gradient-2", "aiFamilies": "ai-families-1", "commissioning": "commissioning-execution-1", "control": "closed-loop-1", "hardwareReliability": "hardware-reliability-1", "foundations": "foundations-1", "modelLifecycle": "causal-decoder-lifecycle-1", "automation": "atlas-automation-1", "machineExecution": "machine-execution-1" }, "builtAt": "2026-09-07T21:51:56.182Z" }, "lesson": "training-data", "studio": "data", "observation": "Train-only learned byte pair merges and causal row31, same sampled document, response mask boundary; separate byte-BPE specimen from fixed decoder vocabulary.", "start": 12, "end": 18, "duration": 6, "chapter": "03 / DATA", "title": "Boundaries before batches", "caption": "Trace source families, train-only token merges and the causal targets that a packed batch exposes.", "transcript": "The corpus experience fits byte-pair merges only on retained training fields. Inspect a sampled document, response mask and causal key row. Family splits and duplicate diagnostics expose leakage. This byte-BPE specimen is separate from the decoder’s fixed nine-token vocabulary." }, { "file": "03-learning.png", "capturedAt": "2026-09-07T21:56:17.499Z", "sha256": "798aca9cfd633d94957c3dbe7da878369d2dc58b9fddb530e01f3e5ab331ac03", "captureBuild": { "schemaVersion": 1, "edition": "1.2.0", "commit": "b3c87e188a41f68def59031e39c8f89b30f4ad95", "dirty": true, "sourceSha256": "347aadc4a9ef25b674c300c5ffede5a79147692e360a5aaad5411894049dd67a", "contentSha256": "88dcced3116ee6cf691d8d944adb98c78ffc108eb87720bd8224b600e8934749", "modelVersions": { "investigations": "atlas-investigations-1", "processes": "process-mechanisms-1", "learningPrompts": "investigation-prompts-1", "architecture": "architecture-execution-1", "fabrication": "atlas-fabrication-flow-1", "operatingHall": "1", "realization": "atlas-realization-1", "memory": "atlas-memory-interfaces-1", "sensor": "sensor-logistic-1", "retrieval": "lexical-cited-answer-1", "orbital": "1", "progress": "progress-evidence-2", "studioRecords": "atlas-studio-record-1", "memoryWorkload": "memory-workload-bridge-1", "factory": "factory-actual-checkpoint-1", "dataLineage": "data-lineage-1", "evaluation": "identified-evaluation-1", "applicationDecisions": "application-decisions-1", "processCell": "process-cell-acceptance-1", "speculativeVerification": "authored-draft-target-greedy-1", "serving": "causal-kv-serving-1", "learnedRetrieval": "atlas-incident-learned-1", "distributedLearning": "distributed-causal-gradient-2", "aiFamilies": "ai-families-1", "commissioning": "commissioning-execution-1", "control": "closed-loop-1", "hardwareReliability": "hardware-reliability-1", "foundations": "foundations-1", "modelLifecycle": "causal-decoder-lifecycle-1", "automation": "atlas-automation-1", "machineExecution": "machine-execution-1" }, "builtAt": "2026-09-07T21:51:56.182Z" }, "lesson": "tiny-causal-decoder", "studio": "learning", "observation": "Actual seeded 320-update decoder 5fa84f48; copy development8/8, swap0/8; answer-position causal attention and NLL inspected; final evaluation unopened.", "start": 18, "end": 24, "duration": 6, "chapter": "04 / LEARNING", "title": "Weights that actually learn", "caption": "Train the bounded decoder. Inspect its changed attention, target probabilities and real gradients.", "transcript": "A seeded 423-parameter causal decoder completes 320 numerical updates. Copy-development answers reach eight of eight while the deliberately conflicting swap task reaches zero of eight. The selected answer-position attention is causal; the final prompt families remain unopened. Development feedback is not an untouched final test." }, { "file": "04-distributed.png", "capturedAt": "2026-09-07T21:53:43.443Z", "sha256": "3dd2c00b3241adba28a562f48112b639772d5b866427730e2f50cf9ad8c9be99", "captureBuild": { "schemaVersion": 1, "edition": "1.2.0", "commit": "b3c87e188a41f68def59031e39c8f89b30f4ad95", "dirty": true, "sourceSha256": "347aadc4a9ef25b674c300c5ffede5a79147692e360a5aaad5411894049dd67a", "contentSha256": "88dcced3116ee6cf691d8d944adb98c78ffc108eb87720bd8224b600e8934749", "modelVersions": { "investigations": "atlas-investigations-1", "processes": "process-mechanisms-1", "learningPrompts": "investigation-prompts-1", "architecture": "architecture-execution-1", "fabrication": "atlas-fabrication-flow-1", "operatingHall": "1", "realization": "atlas-realization-1", "memory": "atlas-memory-interfaces-1", "sensor": "sensor-logistic-1", "retrieval": "lexical-cited-answer-1", "orbital": "1", "progress": "progress-evidence-2", "studioRecords": "atlas-studio-record-1", "memoryWorkload": "memory-workload-bridge-1", "factory": "factory-actual-checkpoint-1", "dataLineage": "data-lineage-1", "evaluation": "identified-evaluation-1", "applicationDecisions": "application-decisions-1", "processCell": "process-cell-acceptance-1", "speculativeVerification": "authored-draft-target-greedy-1", "serving": "causal-kv-serving-1", "learnedRetrieval": "atlas-incident-learned-1", "distributedLearning": "distributed-causal-gradient-2", "aiFamilies": "ai-families-1", "commissioning": "commissioning-execution-1", "control": "closed-loop-1", "hardwareReliability": "hardware-reliability-1", "foundations": "foundations-1", "modelLifecycle": "causal-decoder-lifecycle-1", "automation": "atlas-automation-1", "machineExecution": "machine-execution-1" }, "builtAt": "2026-09-07T21:51:56.182Z" }, "lesson": "distributed-gradient-weighting", "studio": "distributed", "observation": "Two owned slices of 212 and 211 parameters, one padded zero; 30 targets; one committed update; every replica reconstructs 423 weights and gradients; 1e-9 centralized agreement. Partial shard barrier disabled before second owner.", "start": 24, "end": 30, "duration": 6, "chapter": "05 / DISTRIBUTION", "title": "One objective, many owners", "caption": "Unequal examples need the right denominator. Complete every owned shard before committing an update.", "transcript": "Two ranks contribute gradients for six and twenty-four targets. The 423 parameters use owned slices of 212 and 211, with one unused padded slot. One received shard leaves weights frozen. Once both arrive, owner updates and all-gather reconstruct the same model within the declared centralized-reference tolerance. The depicted run starts at the stated initialization, independently of the prior trained capture." }, { "file": "05-serving.png", "capturedAt": "2026-09-07T21:54:13.800Z", "sha256": "ae2e7b2ba850f4afada35ffb03f4feef6de9120b6d985e9473aafc48457d8e30", "captureBuild": { "schemaVersion": 1, "edition": "1.2.0", "commit": "b3c87e188a41f68def59031e39c8f89b30f4ad95", "dirty": true, "sourceSha256": "347aadc4a9ef25b674c300c5ffede5a79147692e360a5aaad5411894049dd67a", "contentSha256": "88dcced3116ee6cf691d8d944adb98c78ffc108eb87720bd8224b600e8934749", "modelVersions": { "investigations": "atlas-investigations-1", "processes": "process-mechanisms-1", "learningPrompts": "investigation-prompts-1", "architecture": "architecture-execution-1", "fabrication": "atlas-fabrication-flow-1", "operatingHall": "1", "realization": "atlas-realization-1", "memory": "atlas-memory-interfaces-1", "sensor": "sensor-logistic-1", "retrieval": "lexical-cited-answer-1", "orbital": "1", "progress": "progress-evidence-2", "studioRecords": "atlas-studio-record-1", "memoryWorkload": "memory-workload-bridge-1", "factory": "factory-actual-checkpoint-1", "dataLineage": "data-lineage-1", "evaluation": "identified-evaluation-1", "applicationDecisions": "application-decisions-1", "processCell": "process-cell-acceptance-1", "speculativeVerification": "authored-draft-target-greedy-1", "serving": "causal-kv-serving-1", "learnedRetrieval": "atlas-incident-learned-1", "distributedLearning": "distributed-causal-gradient-2", "aiFamilies": "ai-families-1", "commissioning": "commissioning-execution-1", "control": "closed-loop-1", "hardwareReliability": "hardware-reliability-1", "foundations": "foundations-1", "modelLifecycle": "causal-decoder-lifecycle-1", "automation": "atlas-automation-1", "machineExecution": "machine-execution-1" }, "builtAt": "2026-09-07T21:51:56.182Z" }, "lesson": "speculative-decoding-verification", "studio": "serving", "observation": "Actual checkpoint 9ef0dfee carried from distributed update; authored a,EOS rejects at position zero; target output v,b equals independent target greedy; one tentative KV row discarded.", "start": 30, "end": 36, "duration": 6, "chapter": "06 / SERVING", "title": "Verify before accepting", "caption": "Reuse computed keys and values. Reject a wrong proposed prefix and rebuild only the valid target state.", "transcript": "The distributed checkpoint enters the serving queue with the same weight identity. Two learner-authored draft tokens fail target-greedy verification; the target emits v then b, matching its independent greedy reference. One dependent tentative KV row is discarded. This is exact bounded greedy verification, not a learned draft model or measured acceleration." }, { "file": "06-retrieval.png", "capturedAt": "2026-09-07T21:59:57.944Z", "sha256": "fbc45d28a47cd177ef30198507bbb9b6d079d588951a70d6ef608da868078295", "captureBuild": { "schemaVersion": 1, "edition": "1.2.0", "commit": "b3c87e188a41f68def59031e39c8f89b30f4ad95", "dirty": true, "sourceSha256": "347aadc4a9ef25b674c300c5ffede5a79147692e360a5aaad5411894049dd67a", "contentSha256": "88dcced3116ee6cf691d8d944adb98c78ffc108eb87720bd8224b600e8934749", "modelVersions": { "investigations": "atlas-investigations-1", "processes": "process-mechanisms-1", "learningPrompts": "investigation-prompts-1", "architecture": "architecture-execution-1", "fabrication": "atlas-fabrication-flow-1", "operatingHall": "1", "realization": "atlas-realization-1", "memory": "atlas-memory-interfaces-1", "sensor": "sensor-logistic-1", "retrieval": "lexical-cited-answer-1", "orbital": "1", "progress": "progress-evidence-2", "studioRecords": "atlas-studio-record-1", "memoryWorkload": "memory-workload-bridge-1", "factory": "factory-actual-checkpoint-1", "dataLineage": "data-lineage-1", "evaluation": "identified-evaluation-1", "applicationDecisions": "application-decisions-1", "processCell": "process-cell-acceptance-1", "speculativeVerification": "authored-draft-target-greedy-1", "serving": "causal-kv-serving-1", "learnedRetrieval": "atlas-incident-learned-1", "distributedLearning": "distributed-causal-gradient-2", "aiFamilies": "ai-families-1", "commissioning": "commissioning-execution-1", "control": "closed-loop-1", "hardwareReliability": "hardware-reliability-1", "foundations": "foundations-1", "modelLifecycle": "causal-decoder-lifecycle-1", "automation": "atlas-automation-1", "machineExecution": "machine-execution-1" }, "builtAt": "2026-09-07T21:51:56.182Z" }, "lesson": "tool-loop", "studio": "retrieval", "observation": "Frozen six-dimensional fitted retrieval and conditional token decoder; current FAN-7 manual supports generated five-token reservation; same proposal eligible for independently checked fictional runtime.", "start": 36, "end": 42, "duration": 6, "chapter": "07 / KNOWLEDGE", "title": "A reference becomes a proposal", "caption": "Fitted embeddings select context. A small conditional decoder generates a typed reservation proposal.", "transcript": "A separate fitted incident model ranks two fictional manuals, selects the FAN-7 reference and generates reserve FAN-7 FILTER-A 2 followed by EOS. Retrieval score, source support and task correctness have different meanings. The generated proposal carries its frozen model and corpus identity into the workflow." }, { "file": "07-automation.png", "capturedAt": "2026-09-07T22:00:50.365Z", "sha256": "160ac0d88de01f34dab7011c6fcfeff33810cb6ae8f10d46f636ff0f7dfa516c", "captureBuild": { "schemaVersion": 1, "edition": "1.2.0", "commit": "b3c87e188a41f68def59031e39c8f89b30f4ad95", "dirty": true, "sourceSha256": "347aadc4a9ef25b674c300c5ffede5a79147692e360a5aaad5411894049dd67a", "contentSha256": "88dcced3116ee6cf691d8d944adb98c78ffc108eb87720bd8224b600e8934749", "modelVersions": { "investigations": "atlas-investigations-1", "processes": "process-mechanisms-1", "learningPrompts": "investigation-prompts-1", "architecture": "architecture-execution-1", "fabrication": "atlas-fabrication-flow-1", "operatingHall": "1", "realization": "atlas-realization-1", "memory": "atlas-memory-interfaces-1", "sensor": "sensor-logistic-1", "retrieval": "lexical-cited-answer-1", "orbital": "1", "progress": "progress-evidence-2", "studioRecords": "atlas-studio-record-1", "memoryWorkload": "memory-workload-bridge-1", "factory": "factory-actual-checkpoint-1", "dataLineage": "data-lineage-1", "evaluation": "identified-evaluation-1", "applicationDecisions": "application-decisions-1", "processCell": "process-cell-acceptance-1", "speculativeVerification": "authored-draft-target-greedy-1", "serving": "causal-kv-serving-1", "learnedRetrieval": "atlas-incident-learned-1", "distributedLearning": "distributed-causal-gradient-2", "aiFamilies": "ai-families-1", "commissioning": "commissioning-execution-1", "control": "closed-loop-1", "hardwareReliability": "hardware-reliability-1", "foundations": "foundations-1", "modelLifecycle": "causal-decoder-lifecycle-1", "automation": "atlas-automation-1", "machineExecution": "machine-execution-1" }, "builtAt": "2026-09-07T21:51:56.182Z" }, "lesson": "tool-loop", "studio": "automation", "observation": "Same fitted proposal enters fictional lost-reply case, exact UI reservation approved by agent for QA, recovery yields1reservation1workorder with inventory conserved,5calls1error2effects;560syntheticms.", "start": 42, "end": 48, "duration": 6, "chapter": "08 / ACTION", "title": "Inspect what really happened", "caption": "A lost reply can hide a committed write. Recover its receipt and check the actual environment.", "transcript": "The same learned proposal enters a fictional maintenance workflow. After exact reservation review, a reply is deliberately lost. Durable recovery completes one reservation and one linked work order, conserves inventory and avoids duplicate effects. Independent environment checks establish this bounded task result." }, { "file": "08-control.png", "capturedAt": "2026-09-07T22:01:59.792Z", "sha256": "44b351c8f9d96db6804d71f339709d4dd593537c361bd91e89aebe4758f7ab63", "captureBuild": { "schemaVersion": 1, "edition": "1.2.0", "commit": "b3c87e188a41f68def59031e39c8f89b30f4ad95", "dirty": true, "sourceSha256": "347aadc4a9ef25b674c300c5ffede5a79147692e360a5aaad5411894049dd67a", "contentSha256": "88dcced3116ee6cf691d8d944adb98c78ffc108eb87720bd8224b600e8934749", "modelVersions": { "investigations": "atlas-investigations-1", "processes": "process-mechanisms-1", "learningPrompts": "investigation-prompts-1", "architecture": "architecture-execution-1", "fabrication": "atlas-fabrication-flow-1", "operatingHall": "1", "realization": "atlas-realization-1", "memory": "atlas-memory-interfaces-1", "sensor": "sensor-logistic-1", "retrieval": "lexical-cited-answer-1", "orbital": "1", "progress": "progress-evidence-2", "studioRecords": "atlas-studio-record-1", "memoryWorkload": "memory-workload-bridge-1", "factory": "factory-actual-checkpoint-1", "dataLineage": "data-lineage-1", "evaluation": "identified-evaluation-1", "applicationDecisions": "application-decisions-1", "processCell": "process-cell-acceptance-1", "speculativeVerification": "authored-draft-target-greedy-1", "serving": "causal-kv-serving-1", "learnedRetrieval": "atlas-incident-learned-1", "distributedLearning": "distributed-causal-gradient-2", "aiFamilies": "ai-families-1", "commissioning": "commissioning-execution-1", "control": "closed-loop-1", "hardwareReliability": "hardware-reliability-1", "foundations": "foundations-1", "modelLifecycle": "causal-decoder-lifecycle-1", "automation": "atlas-automation-1", "machineExecution": "machine-execution-1" }, "builtAt": "2026-09-07T21:51:56.182Z" }, "lesson": "robot-feedback-control", "studio": "control", "observation": "Fitted0.9 retention/0.2 gain;81 candidate four-step futures; actual0.2s discrete plant action; default final error0.0006m; task constraints independently checked; action effort not energy.", "start": 48, "end": 54, "duration": 6, "chapter": "09 / CONTROL", "title": "Prediction meets consequence", "caption": "Fit dynamics, inspect candidate futures and compare the chosen action with the state that follows.", "transcript": "Twelve calibration transitions fit the plant model. The controller evaluates eighty-one four-step action sequences, then applies one action to a separate simulated trajectory. Observation delay, noise, model error and deadlines can change the outcome. Action effort is not physical energy; no robot trial is implied." }, { "file": "09-factory.png", "capturedAt": "2026-09-07T22:04:32.540Z", "sha256": "b25f9bb8c0f22891ab5bb5aa76f28bc19111399770e621305bfac0060c7a18bf", "captureBuild": { "schemaVersion": 1, "edition": "1.2.0", "commit": "b3c87e188a41f68def59031e39c8f89b30f4ad95", "dirty": true, "sourceSha256": "d2548353ee9954440987154bfe53a3c6551ce8ddfdadabf8e0a51a469602e44b", "contentSha256": "ef415c169225037971ccaee7ed4f02b821a3485786afe9bf1037029a9fb06111", "modelVersions": { "investigations": "atlas-investigations-1", "processes": "process-mechanisms-1", "learningPrompts": "investigation-prompts-1", "architecture": "architecture-execution-1", "fabrication": "atlas-fabrication-flow-1", "operatingHall": "1", "realization": "atlas-realization-1", "memory": "atlas-memory-interfaces-1", "sensor": "sensor-logistic-1", "retrieval": "lexical-cited-answer-1", "orbital": "1", "progress": "progress-evidence-2", "studioRecords": "atlas-studio-record-1", "ranking": "exposure-aware-ranking-1", "memoryWorkload": "memory-workload-bridge-1", "factory": "factory-actual-checkpoint-1", "dataLineage": "data-lineage-1", "evaluation": "identified-evaluation-1", "applicationDecisions": "application-decisions-1", "processCell": "process-cell-acceptance-1", "speculativeVerification": "authored-draft-target-greedy-1", "serving": "causal-kv-serving-1", "learnedRetrieval": "atlas-incident-learned-1", "distributedLearning": "distributed-causal-gradient-2", "aiFamilies": "ai-families-1", "commissioning": "commissioning-execution-1", "control": "closed-loop-1", "hardwareReliability": "hardware-reliability-1", "foundations": "foundations-1", "modelLifecycle": "causal-decoder-lifecycle-1", "automation": "atlas-automation-1", "machineExecution": "machine-execution-1" }, "builtAt": "2026-09-07T22:02:24.958Z" }, "lesson": "checkpoint-retained-work", "studio": "factory", "observation": "Reproduced source5fa84f48 optimizer320 carried to factory;5executed/2rolledback/3durable, optimizer323 identity0bd56b4e, RNG1313532916 and12additionaldraws;30.8785assumeds,2981.789facilityJ. Committed state matches uninterrupted reference.", "start": 54, "end": 60, "duration": 6, "chapter": "10 / INFRASTRUCTURE", "title": "Keep the work that survives", "caption": "Carry learned weights through admission, a partial checkpoint write, rollback and durable recovery.", "transcript": "A reproduced 320-update decoder enters the factory job. Five additional updates execute; two roll back and three remain durable. The accepted optimizer reaches step 323 with matched weights, moments, RNG and cursor relative to uninterrupted work. Time and facility energy use declared synthetic assumptions. Return to the atlas to change a boundary and explain the result." } ], "music": "Open Horizons: original procedural AI Atlas instrumental, 128 BPM. No sampled song or recording. Original score and notices preserved in examples/walkthrough.", "artifactSha256": "05ba66f3d933ff9ca1a5b7bbd94a6aaa0b30d7f708991d1f7347cf6b37d365ba" }