reaperdoesntknow/Symiotic-14B

TEXT GENERATIONPricing:Input $0.48 / Output $0.96Concurrent Unit Cost:1Model Size:14BQuant:FP8Context Size:32kTool Calling:SupportedPublished:May 6, 2025License:afl-3.0Architecture:Transformer0.0K Featherless Exclusive Cold

reaperdoesntknow/Symiotic-14B is a 17.8-billion-parameter symbolic-transformer hybrid model, built on the Qwen-14B backbone, designed for advanced cognitive reasoning. It integrates neural representation with structured symbolic cognition, featuring persistent memory, symbolic routing, and self-organizing knowledge structures. This experimental model excels in multi-step reasoning, symbolic math/code generation, and scientific dialogue within fuzzy problem domains.

Loading preview...

Overview

reaperdoesntknow/Symiotic-14B is a 17.8-billion-parameter symbolic-transformer hybrid model, developed by reaperdoesntknow (Convergent Intelligence LLC: Research Division). Built on the Qwen-14B backbone, it combines high-capacity neural representation with structured symbolic cognition, aiming for full-scale cognitive reasoning. This experimental research checkpoint focuses on architectural intent rather than benchmarked results, offering unique capabilities for complex problem-solving.

Key Capabilities & Architectural Highlights

  • Hybrid Architecture: Integrates a Qwen-14B transformer with specialized symbolic modules like ThoughtDynamicsLNN, LiquidThoughtProcessor, and CrystallineProcessor.
  • Persistent Memory: Features 4096 symbolic states in FP32, retrieved using entropy and contextual similarity, enabling true memory in conversational agents.
  • Symbolic Cognition: Supports multi-stage symbolic routing, self-organizing knowledge structures, and a "Dream Mode" for background symbolic simulation.
  • Mathematical Foundations: Connects to Discrepancy Calculus (DISC) for self-generating completeness and memory consolidation stability.

Intended Use Cases

  • Advanced Reasoning Agents: Ideal for multi-step conversational agents requiring true memory.
  • Symbolic Generation: Excels in long-form symbolic theorem generation, proof planning, and symbolic math/code synthesis.
  • Scientific & Fuzzy Domains: Suitable for scientific dialogue, symbolic simulations, and reasoning in discontinuous or non-smooth problem domains.

Limitations

  • Memory utility requires curation and seeding.
  • Symbolic cognition is not instruction-tuned for general QA.
  • Increased VRAM usage during generation due to FlashAttention and symbolic modules.

Finetunes
1 models