README.md

metadata

title: SRT-NLA v1 demo
emoji: 🪞
colorFrom: indigo
colorTo: green
sdk: gradio
sdk_version: 4.44.1
python_version: 3.10.13
app_file: app.py
pinned: false
license: apache-2.0
hardware: zero-a10g
short_description: Latent autoencoder over Qwen2.5-7B L20
models:
  - Qwen/Qwen2.5-7B
  - RiverRider/srt-nla-av-v1
tags:
  - srt
  - semiotic-reflexive-transformer
  - interpretability
  - introspection
  - uncertainty
  - visualization
  - llm
thumbnail: >-
  https://huggingface.co/spaces/RiverRider/srt-nla-av-v1-demo/resolve/main/thumbnail.png

SRT-NLA v1 — latent autoencoder demo

Two interactive views of the RiverRider/srt-nla-av-v1 verbalizer.

1. Round-trip autoencoder. Encode a passage with frozen Qwen2.5-7B at L20, decode the resulting 3584-dim vector with the 12.7M-parameter NLA verbalizer, and re-encode the verbalization to measure round-trip fidelity (centered fve_nrm, rho_norm). Best-of-N rerank is included — this is the technique that takes greedy ρ ≈ 0.26 up to ρ ≈ 0.92 in the paper.
2. Latent arithmetic. Encode two passages, slide an interpolation α, and verbalize the mixture vector.

Anchors (paper §3)

Anchor	centered fve_nrm
Random floor	0.510
NN-retrieval over pool=2000	0.71
Qwen paraphrase ceiling	0.799

rho_norm = (cen − 0.510) / 0.289 linearly remaps to [0, 1] against these anchors.

Source

Code: https://github.com/space-bacon/SRT (branch nla, docs/hf/nla_v1_demo/).