公开知识图谱记录

NeoMME-Retriever

记录摘要

Dual-head design: dense head uses mean pooling, late-interaction head for fine-g

来自项目知识图谱的有来源实体；具体工具属性仍需补齐。

基准测试: ViDoRe v3 Pareto frontier for nDCG@10 and model size
基准测试: 0.523 nDCG@10 (260M) and 0.556 (800M) on ViDoRe v3
基准测试: 0.523 nDCG@10 on ViDoRe v3
基准测试: 0.556 nDCG@10 on ViDoRe v3
基准测试: ViDoRe v3
能力: returns dense and late-interaction embeddings in one forward pass
能力: ranks document page screenshots directly, bypassing OCR
能力: Retrieval-finetuned using mean-pooled dense head and late-interaction head

知识图谱当前记录了 8 个已知资料维度，另有 10 个维度尚未记录，关联 2 篇公开资料。这些数字只描述资料覆盖，不代表模型性能或适配性已经验证。