TheMindExpansionNetwork/mindbotz-reality-forge-the-one-v1

Mindbotz Reality Forge: The One v1

Private launch-candidate dataset for the first serious Mind Expansion Forge / Spark Expansion run.

The One v1 consolidates the larger local synthetic corpora into one Hugging Face-ready dataset while preserving the Reality Forge v0.2 tokens and HY-World world-eyes lane.

YOLO omni text+vision config

Use the_one_omni for the all-with-eyes Spark Expansion. It mixes text/team behavior and world-eyes image rows into one VLM training lane.

• Train: 3347
• Validation: 430
• Test: 391

Recommended training config

Use the_one_text_unique first. It removes exact repeated chat transcripts while preserving the larger MindBot Ultra set and one copy of each Team Office pattern.

• Unique text train: 3220
• Unique text validation: 414
• Held-out text eval/test: 375

Optional curriculum config

Use the_one_text_full only if we intentionally want repetition/weighting from the source dataset.

• Full text train: 6025
• Full text validation: 639

World-eyes config

Keep world_eyes_vision separate until the multimodal Gemma/processor path is verified.

• Vision train: 127
• Vision validation: 16
• Vision test: 16

Source lanes

• Jimsky Team Office unified 15-agent dataset: command center, routing, handoffs, safety gates.
• MindBot Ultra Synergistic Cognition 3333: synthetic reasoning/persona/framework lane.
• Reality Forge v0.2 seed: realm tokens and expansion map overlay.
• HY-World / Reality Forge world-eyes: prompt-to-world image/text pairs.

Safety / scope

Private dataset. Training has not started from this upload. Do not use examples to claim proven AI consciousness; Mind Expansion is creative language for synthetic training and adapter development.