主题分类/UnsortedRESEARCH LIBRARYUnsorted新收录但尚未归类的论文。全部标签agentic-aiagentsalgebraic-complexityattentionattributionbrain-encodingchain-of-thoughtcompressioncontext-compactioncopy-mechanismdata-augmentationdecision-modelsdomain-adaptationefficiencyefficient-inferenceembeddingsensemblefactual-knowledgefine-tuningfmrigeneralizationgradient-bottleneckhashinghyper-connectionsinferenceinterpretabilitylanguage-modelinglanguage-modelslearning-theorylinear-probesllmllm-evaluationlogic-programminglong-contextlooped-transformersmoemultilingualmultimodalneurosymbolicoption-cachingparametric-memoryparitypersistent-statephase-aware-attentionprivacyreasoningrecurrencerecurrent-depthreinforcement-learningresidual-streamretrievalrobustnessselective-readingset-functionsspectral-optimizationspeculative-decodingstate-trackingstructured-layerstensor-algebratokenizationtraining-efficiencytransformerunlearningvocabulary-expansion共 21 篇论文 · 按研读更新时间排序2609.32814v1 · r1AlgebraLayersChange the Product, Keep the Parameters: Associative Algebra Layers for Transformers结合代数通过保留权重、减少固定交互降低投影计算;小模型生成提速约7%但质量下降,理论与内核证据需和递归模型实验分开理解。transformerstructured-layersalgebraic-complexityefficient-inference阅读笔记汇报 PDF ↗论文原文 ↗没有找到匹配论文,试试其他关键词或标签。← 上一页第 2 / 2 页分页与侧边栏链接可直接使用;关键词筛选和全文搜索需要启用 JavaScript。
2609.32814v1 · r1AlgebraLayersChange the Product, Keep the Parameters: Associative Algebra Layers for Transformers结合代数通过保留权重、减少固定交互降低投影计算;小模型生成提速约7%但质量下降,理论与内核证据需和递归模型实验分开理解。transformerstructured-layersalgebraic-complexityefficient-inference阅读笔记汇报 PDF ↗论文原文 ↗