#knowledge injection

6 stories · 0 in 7 days · 1 in 30 days

DeepSeek-V4.1-Flash 1 DKL 1 GPT-2 1 Instruct LLM 1 Qwen3.8-Flash 1 RAG 1 WOS 1 large language models 1

Stories, products and papers tagged "knowledge injection", newest first.

Oct 16

Papers BuildersMulti-source09:17

Chinese content研究测试向 Qwen3.8-Flash 与 DeepSeek-V4.1 记忆模块写入知识的可行性

把知识写进模型记忆到底行不行？这篇论文在 Qwen 和 DeepSeek 上做了实测，命中率数据挺扎眼，做记忆机制研究的值得看数据。

#Qwen3.8-Flash #DeepSeek-V4.1-Flash #GPT-2 #knowledge injection

Event

arXiv: DeepSeek@Wei Gao et al. (4)Source

Sep 3

PapersSelected73°12:14

DKL method for instruction-tuned language models

DKL avoids expensive instruction fine-tuning while preserving instruction-following capabilities, making it more practical for real-world applications.

#DKL #RAG #Instruct LLM #knowledge injection

arXiv cs.AI@Kushagra Bhushan et al. (10)Source

Sep 2

PapersSelected73°10:47

Confusion-Aware Retrieval Framework for Text Classification

The approach improves text classification accuracy for smaller models, making it cost-effective for businesses across ASEAN with limited computational resources.

#text classification #large language models #RAG #knowledge injection

arXiv cs.AI@Manish Gupta et al. (3)Source

Jul 23

Papers13:07

Chinese content无梯度下降，将知识直接写入Transformer块的新方法

斯坦福团队搞了个新招：不用梯度下降，直接把事实写进Transformer的MLP里，论文被COLM 2026收了，想了解怎么不训练就灌知识的可以看看。

#Transformer #MLP #knowledge injection #COLM 2026