🏗️ Building on HF

Burton Lancaster PRO

RiverRider

3 2 16

https://sunstonenorth.com

Space-Bacon

AI & ML interests

Explainable AI

Recent Activity

updated a model about 5 hours ago

RiverRider/srt-sunstone-linear-head

repliedto their post about 5 hours ago

Train Once, Read Everywhere Paper title: Train Once, Read Everywhere: Substrate Invariance of the Linearly Readable Structure in Frozen Language Models Paper URL: https://github.com/space-bacon/SRT/blob/main/arxiv_program/paper.md Repository URL: https://github.com/space-bacon/SRT The consolidated findings of the SRT research program are now available. The program treats frozen production-scale language models as substrates whose internal states carry structure that small, inspectable instruments can read. Results include: - A ~12 M-parameter adapter that surfaces per-token semiotic signals from a frozen 7 B backbone with zero cross-entropy degradation - An activation verbalizer that recovers text from single hidden states up to a calibrated paraphrase ceiling - Linear readout ports spanning dense 3 B models to 94-layer 235 B mixture-of-experts models - A 22 MB linear head that gives a frozen multimodal chat model image-to-text retrieval performance matching fully trained 2018 dual encoders on the COCO benchmark The central claim is substrate invariance. The readable structure is a stable property of the model class. A head trained once on one host reads, with no retraining and at most a 42 KB recalibration, across: - Hosts ten times smaller (31 B → 3 B) - 4-bit weight precision - Entirely different silicon and kernels (CUDA/bf16 to Apple Silicon/MLX-Q4) Deployment tiers differ in latency and cost, never in capability. All instruments, measurement protocols, invariance evidence, negative results, and artifacts are in the repository.

updated a model about 5 hours ago

RiverRider/srt-nla-gemma4-artifacts

View all activity

Organizations

liked a dataset 15 days ago

dipankarsarkar/gpuemu-corpus

Viewer • Updated Jun 27 • 26 • 119 • 2

liked a Space 15 days ago

The Correctness Illusion

🧮

Detect hidden bugs in GPU kernels with differential fuzzing

liked a model about 1 month ago

thinkingmachines/Inkling

Image-Text-to-Text • 952B • Updated Jul 23 • 152k • • 1.74k

liked 3 Spaces about 2 months ago

zooL4nD3r demo

🦓

Translate a passage across 961 discourse communities

Gemma-4-31B-it SRT-Sunstone

🔮

A read-out that reads images — trained only on words

Metacognition Leaderboard

🧠

Explore LLM metacognition rankings and submit your model

liked 2 Spaces 2 months ago

SRT Showcase

🧭

Watch a frozen Qwen-2.5-7B think — live SRT introspection

Semantic Compression

💬

Compress text while preserving meaning

liked 2 Spaces 3 months ago

SRT introspect

🧭

Adaptive-density reasoning traces over a frozen Qwen-2.5-7B

MindReader-NLA

🧠

Ask a frozen LM what it is thinking, in plain English.

liked a model 3 months ago

RiverRider/srt-nla-av-v1

Feature Extraction • Updated Jun 18 • 18 • 4

liked a Space 3 months ago

CanIRunThisLLM

🎨

Estimate VRAM needs and speed for LLMs on any GPU

liked 2 Spaces 4 months ago

Contrastive Steering

🌓

Steer the model outputs with an activation delta.

De-Generate

🤖

Detect LLM generated text.

liked a model 4 months ago

RiverRider/srt-adapter-v8a

Feature Extraction • Updated Jun 18 • 28 • 2

liked a Space 4 months ago

SRT-Adapter v8a Demo

🌊

Per-token reflexivity heatmap from a frozen Qwen2.5-7B