Hy-MT1.5-1.8B-1.25bit

Extreme quantized 1.8B LM on HuggingFace

A single HuggingFace model at rank 73 with 1.25-bit quantization but no cross-platform signal. 178 of 200 pool entries are single-source, making this a common pattern rather than an anomaly.

Why now: HuggingFace scrape captured new upload; no external validation or event triggered visibility.

• Only signal from HuggingFace rank 73; zero presence on GitHub, HN, X, Reddit, Product Hunt, Dev.to, Bluesky
• HF score 79 but normalized to 0.117 — weak within its own source
• 178/200 pool entries are single_source; this is baseline distribution noise, not signal

Considerations: 1.25-bit quantization is technically interesting but the 'AngelSlim' namespace has no known reputation, noGitHub presence, no discussion threads, and no downloads/engagement metrics visible. Could be test upload or SEO-optimized model card. No reproductions, no integrations, no issue activity.

EMERGING SIGNAL · Ignore: No action warranted until cross-source signal emerges or known team attaches credibility.

Sources: HuggingFace: AngelSlim/Hy-MT1.5-1.8B-1.25bit