相关数据集
Mr-Verse1/SEFQ-1bit-LLM
SEFQ（频谱误差反馈量化）是一种将FP16大语言模型转换为1-bit表示的提议方法，无需大规模重训练。其核心思想结合了频谱初始化（保持权重方向）、误差反馈量化（纠正累积损失）和自适应组缩放（保护重要权重）。该方法旨在实现：训练后转换、约28倍内存减少、最小精度损失（目标<10%）、在低资源设备上部署。
Hugging Face2026-05-01 更新90
Relative substitution rates of the ten slowest and fastest changing features in our example analysis of Austronesian typological data.
Relative substitution rates of the ten slowest and fastest changing features in our example analysis of Austronesian typological data.
Figshare2017-08-11 更新50
Average of continuation indexes of characters of same level texts with 2 levels difficulty in training corpus.
Average of continuation indexes of characters of same level texts with 2 levels difficulty in training corpus.
NIAID Data Ecosystem30
Data and code for the article 'Looking for iconicity in Attic oratory: Four quantitative case studies', Rhetorica 42.2
This set contains the data files referred to in the text of the article 'Looking for iconicity in Attic oratory: Four quantitative case studies'. The article specifies which file number is relevant to
DataCite Commons2023-02-16 更新60
Configuration of the LastBERT model.
This work focuses on the efficiency of the knowledge distillation approach in generating a lightweight yet powerful BERT-based model for natural language processing (NLP) applications. After the model
NIAID Data Ecosystem30
