reaperdoesntknow/Symbiotic-1B

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:0.8BQuant:BF16Ctx Length:32kTool Calling:SupportedPublished:May 6, 2025License:afl-3.0Architecture:Transformer0.0K Warm

SymbioticLM-1B by reaperdoesntknow is a 1 billion parameter hybrid symbolic-transformer model, built on a Qwen-1B backbone. It integrates a rotary transformer with a symbolic processing pipeline and a persistent episodic memory for enhanced reasoning. This model is optimized for lightweight, memory-augmented symbolic inference in constrained environments, excelling at tasks like procedural planning, math modeling, and graph-based explanation generation.

Loading preview...

SymbioticLM-1B: A Hybrid Reasoning Model

SymbioticLM-1B, developed by reaperdoesntknow, is a compact 1 billion parameter model that combines a Qwen-1B rotary transformer with a unique symbolic processing pipeline and a 2048-vector episodic memory. This architecture is designed for advanced reasoning in resource-limited settings, leveraging a "cognitive engine" that includes symbolic memory, dynamic thought evolution, and entropy-gated control. It is part of the Convergent Intelligence LLC portfolio, built on the Discrepancy Calculus (DISC) framework, which treats training singularities as structural signals for learning.

Key Capabilities

  • Hybrid Architecture: Fuses neural (Qwen-1B) and symbolic processing for robust reasoning.
  • Memory-Augmented: Utilizes 2048 symbolic vectors with entropic and contextual retrieval for persistent knowledge.
  • Symbolic Modules: Incorporates advanced components like ThoughtDynamicsLNN, CrystallineProcessor (DNAConv GNN), and LiquidThoughtProcessor.
  • Dream Mode: Features a "Dream Mode" for symbolic simulation via a ThoughtGenerator.
  • CPU-Optimized: Designed for efficient inference on CPUs and embedded systems.

Good For

  • Symbolic Reasoning: Excels in tasks requiring logical processing and structured thought.
  • Constrained Environments: Ideal for lightweight assistants and agents where memory and computational resources are limited.
  • Specific Applications: Suitable for educational agents, graph-based explanation generation, procedural planning, mathematical modeling, and small-scale code generation.

Limitations

  • Less fluent in free-form language compared to larger, purely neural models.
  • Symbolic accuracy benefits from memory curation.
  • Complex queries in "Dream Mode" may require warm-up or symbolic seeding.

Finetunes
1 models