We built BSE (Bramble Semantic Engine) – a semantic compressor that transforms natural inputs into low-dimensional structured representations.It's designed as a preprocessing engine for LLMs, capable of reducing long inputs into compact, logic-preserving forms across:1. LanguageExtracts SVO (Subject, Verb, Object) structureCaptures modifiers: adjectives/adverbsRestores pronouns from short-term memoryDetects questionsComputes:Compression Rate (%)Semantic Loss (%)Compares sentence compression outputs via SDC:Subject-Subject, Verb-Verb, Object-Subject similaritySentence distance2. ImageCrops and weights center-priority patchesConverts into 100x100 weighted matricesVisualizes:R, G, B ChannelsBrightness3. AudioDecomposes audio into pitch & intensity across frequency bandsReturns normalized 2D matricesVisualized as grayscale spectro-patchesLive demo (Gradio): https://huggingface.co/spaces/Sibyl-V/BSE_demoFeedback welcome on:Compression logicUse cases (LLM fine-tuning, retrieval, alignment)Design of multi-modal structure outputBuilt in 48 hours by a solo dev & their black nine-tailed fox partner. Let us know what you'd improve — and what scares you.
First seen: 2025-04-13 06:58
Last seen: 2025-04-13 07:59