LCLM
LCLM的通用不是指 训练完成之后我直接换一个 decoder模型就能接着用。 这里的通用实际上指的是: 跨任务、跨 domain、query-agnostic。 论文用 350B tokens 大数据量训练了一套可以理解多领域知识的LCLM， 整个训练好的encoder-decoder是通用领域的。 不是单独指encoder是通用的。
Original Note
LCLM Read Deep
- papers
- Original Note
- Updated: 2026-10-12T10:39:27+08:00
Source provenance
- Canonical route
- /original-notes/papers/Compressor/LCLM/LCLM Read Deep
- Source path
- papers/Compressor/LCLM/LCLM Read Deep.md
- Content hash
- 1c8ba3551d914b4a8d5802faef505603059b4d5108e99e525fe50745fe86539b
- Updated commit
- c9c2d5eccd9ee09ce82352f0570d53e8b75bbb2e
- Updated at
- 2026-10-12T10:39:27+08:00
- Document type
- Original Note
- Source Collection
- papers
- Source Path
- papers/Compressor/LCLM/LCLM Read Deep.md
- Type
- Original Note
- Updated At
- 2026-10-12T10:39:27+08:00
Evidence-backed relations
Source Note · Same Topic
- Source Note · Same Topic SapBERT:
- Source Note · Same Topic 1. To understand list:
- Source Note · Same Topic Evaluation of 9 BEL models Read Deep
- Source Note · Same Topic BELB Read Deep
- Source Note · Same Topic BeLink Read Deep
- Source Note · Same Topic 经典rerank Read Deep
- Source Note · Same Topic BGE-M3 Read Deep
- Source Note · Same Topic BioELX Read Deep
- Source Note · Same Topic BioSyn 论文讲解分析 PPT 大纲
- Source Note · Same Topic Biomedical Entity Representations with Synonym Marginalization
Evidence-backed relations
Related Summary
- Related Summary · Same Topic CBLUE Abstract Summary