DepthOk4115

• Embracing the noise: How to build an agent that is both neuro-symbolic and probabilistic.

  Posted by DepthOk4115@reddit | LocalLLaMA | View on Reddit | 10 comments

• [R] Intrinsic curiosity on text embeddings: a 5-component reward function with developmental annealing, running on real agents

  Posted by DepthOk4115@reddit | LocalLLaMA | View on Reddit | 1 comments

• Sanity check on Milla Jovovich's MemPalace: Mixed metrics, bypassed judges, and that 96.6% LongMemEval score

  Posted by DepthOk4115@reddit | LocalLLaMA | View on Reddit | 42 comments

• Claude Mythos and the "Too Dangerous to Release" Marketing Playbook

  Posted by DepthOk4115@reddit | LocalLLaMA | View on Reddit | 0 comments

• We are duct-taping LLMs to logic engines. Here is why the recent "Awareness-First" paper is my 2026 roadmap for true Neuro-Symbolic Agents.

  Posted by DepthOk4115@reddit | LocalLLaMA | View on Reddit | 1 comments