From e8c747a6dcb3a16078cb5af4efcd8cc47d85f3f8 Mon Sep 17 00:00:00 2001 From: PerryLink Date: Sat, 10 Oct 2026 13:04:59 +0800 Subject: [PATCH] =?UTF-8?q?v1.1=20errata:=20Phocinae=20entry=20figures=20(?= =?UTF-8?q?0.906/0.848=20=C2=B7=2021.0ms(RTX=205090))?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 8fbbcef..d788ab4 100644 --- a/README.md +++ b/README.md @@ -109,7 +109,7 @@ Open weights and code that rebuild the System One shape from encoders, small dec - [jevos](https://github.com/feder-cr/jev), MiniCPM5-1B cut to 17 layers and served as a 619 MB GGUF on CPU-only llama.cpp; answers Noul (yes/no) questions over Jev's `/v1/systemone` wire in 54 to 220 ms, 0.815 accuracy versus Jev's 0.927. [![Code](https://img.shields.io/github/stars/feder-cr/jev?style=flat-square&logo=github&label=Code&color=181717)](https://github.com/feder-cr/jev) - [WebJev](https://github.com/lexmount/WebJev), Apache-2.0 Qwen3.5-35B-A3B fine-tune for browser agents that returns a probability for every operation and element option over Jev's `/v1/systemone` wire, released with its training recipe and live-web training data; authors report 38.52% on 125 real-website tasks graded by deterministic verifiers versus 16.67% for Jev 1.13 inside the same jev-ultrafast agent. [![Code](https://img.shields.io/github/stars/lexmount/WebJev?style=flat-square&logo=github&label=Code&color=181717)](https://github.com/lexmount/WebJev) [![Model](https://img.shields.io/badge/%F0%9F%A4%97%20Model-8B5CF6?style=flat-square)](https://huggingface.co/Lexmount/WebJev-35B-A3B) [![Dataset](https://img.shields.io/badge/%F0%9F%A4%97%20Dataset-06B6D4?style=flat-square)](https://huggingface.co/datasets/Lexmount/WebJev) - [WaterSheep](https://github.com/SamratDuttaOfficial/WaterSheep), Apache-2.0 decision model that returns a calibrated probability for every Noul, Choice, Score and multi-label option and serves Jev's `/v1/systemone` wire; the official Python SDK works against it unchanged. [![Code](https://img.shields.io/github/stars/SamratDuttaOfficial/WaterSheep?style=flat-square&logo=github&label=Code&color=181717)](https://github.com/SamratDuttaOfficial/WaterSheep) [![Model](https://img.shields.io/badge/%F0%9F%A4%97%20Model-8B5CF6?style=flat-square)](https://huggingface.co/samratduttaofficial/WaterSheep) [![Website](https://img.shields.io/badge/Website-2EA44F?style=flat-square&logo=googlechrome&logoColor=white)](https://samratduttaofficial.github.io/WaterSheep/) -- [Phocinae-Largha-150M-v1](https://github.com/Phocinae/Phocinae-Largha-150M-v1), Apache-2.0 bilingual (English/Chinese) decision model on mmBERT-small that returns calibrated Choice, Score and Noul probabilities in one 144M-parameter forward pass; authors report typed-decisions 0.797 (en) / 0.789 (zh) at 18.6 ms p50 on an RTX 5090. [![Code](https://img.shields.io/github/stars/Phocinae/Phocinae-Largha-150M-v1?style=flat-square&logo=github&label=Code&color=181717)](https://github.com/Phocinae/Phocinae-Largha-150M-v1) [![Model](https://img.shields.io/badge/%F0%9F%A4%97%20Model-8B5CF6?style=flat-square)](https://huggingface.co/Phocinae/Phocinae-Largha-150M-v1) +- [Phocinae-Largha-150M-v1](https://github.com/Phocinae/Phocinae-Largha-150M-v1), Apache-2.0 bilingual (English/Chinese) decision model on mmBERT-small that returns calibrated Choice, Score and Noul probabilities in one 144.3M-parameter forward pass; authors report typed-decisions 0.906 (en, specialist: fitted on this dataset's train split) / 0.848 (zh, machine-translated cases) at 21.0 ms p50 on an RTX 5090. [![Code](https://img.shields.io/github/stars/Phocinae/Phocinae-Largha-150M-v1?style=flat-square&logo=github&label=Code&color=181717)](https://github.com/Phocinae/Phocinae-Largha-150M-v1) [![Model](https://img.shields.io/badge/%F0%9F%A4%97%20Model-8B5CF6?style=flat-square)](https://huggingface.co/Phocinae/Phocinae-Largha-150M-v1) ---