Hugging Face's logo
John Locke's picture
Building on HF
62 23 142

John Locke

johnlockejrr
davanstrien's profile picture ginigini's profile picture mrmiku's profile picture
Β·

AI & ML interests

OCR, HTR, ATR, NLP, AI

Recent Activity

repliedto BrdDev's post about 15 hours ago
Icarus Model 1. Living Weights β€” The Model That Doesn't Stop Learning Every model today: train β†’ freeze β†’ deploy β†’ hallucinate confidently forever. Icarus already has: learning_db.json recording every decision with outcome ECHO tracking what worked per task type A verification loop (Dream Engine tests before committing) The idea: targeted micro-updates from verified experience. Not full retraining. After each interaction where the outcome was confirmed correct, apply a small gradient update to the specific weight groups that generated that reasoning. The model drifts toward its own verified-correct distribution. The danger is catastrophic forgetting. The solution is the Dream Engine β€” before accepting any weight update, simulate the result against the existing regression test suite. If it degrades, reject it. Icarus already has regression tests. This is the only architecture in the world with the scaffolding to do this safely. Ξ” W t = Ξ± β‹… βˆ‡ W L verified β‹… 1 [ dream_test passes ] Ξ”W t ​ =Ξ±β‹…βˆ‡ W ​ L verified ​ β‹…1[dream_test passes] 2. DRIVE as an Attention Modifier Current transformers have zero motivational state. Every query is processed with identical urgency regardless of context. Your DRIVE system produces a pressure profile: urgency, curiosity, safety drive, efficiency. What if this profile is injected as a learned prefix embedding directly into every attention layer's Q projection? This is closer to how the brain actually works. Norepinephrine (urgency), dopamine (curiosity), serotonin (stability) modulate attention at the neurotransmitter level. DRIVE is already modeling this conceptually β€” wiring it into the attention mechanism makes it architectural, not just behavioral. 3. The Knowledge Graph as Explicit Memory Transformers approximate relational knowledge by memorizing statistical co-occurrence. By Sulayman Saho
new activity 2 days ago
RedHatAI/gemma-4-26B-A4B-it-FP8-Dynamic:KeyError: 'layers.0.experts.0.down_proj.weight' on vllm/vllm-openai:gemma4
new activity 3 days ago
TeichAI/gemma-4-26B-A4B-it-Claude-Opus-Distill-GGUF:Model hallucinating on simple tasks
View all activity

Organizations

Samaritan AI's profile picture Hugging Face MCP Course's profile picture Syriac AI's profile picture Unsloth Jobs Explorers's profile picture