
LLM 论文里程碑
全图 · LLM 论文里程碑 24 / 100 篇已发
14 章 100 篇,从原作的问题、方法与证据理解现代 LLM 体系怎样形成
一、统计学习与神经语言模型
8 / 8
- 论文里程碑 #1:Learning Representations by Back-Propagating Errors
- 论文里程碑 #2:A Statistical Approach to Machine Translation
- 论文里程碑 #3:Long Short-Term Memory
- 论文里程碑 #4:Conditional Random Fields
- 论文里程碑 #5:BLEU
- 论文里程碑 #6:Latent Dirichlet Allocation
- 论文里程碑 #7:A Neural Probabilistic Language Model
- 论文里程碑 #8:Natural Language Processing (Almost) from Scratch
二、表示、序列与训练基础
12 / 12
- 论文里程碑 #9:Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation
- 论文里程碑 #10:Efficient Estimation of Word Representations in Vector Space
- 论文里程碑 #11:Distributed Representations of Words and Phrases and Their Compositionality
- 论文里程碑 #12:Recursive Deep Models for Semantic Compositionality Over a Sentiment Treebank
- 论文里程碑 #13:Dropout: A Simple Way to Prevent Neural Networks from Overfitting
- 论文里程碑 #14:Sequence to Sequence Learning with Neural Networks
- 论文里程碑 #15:Convolutional Neural Networks for Sentence Classification
- 论文里程碑 #16:GloVe: Global Vectors for Word Representation
- 论文里程碑 #17:Neural Machine Translation by Jointly Learning to Align and Translate
- 论文里程碑 #18:Effective Approaches to Attention-based Neural Machine Translation
- 论文里程碑 #19:Adam: A Method for Stochastic Optimization
- 论文里程碑 #20:Layer Normalization
三、子词接口与 Transformer
4 / 4
四、预训练与迁移
0 / 9
五、规模与训练数据
0 / 10
六、架构与长上下文
0 / 7
七、能力评测
0 / 8
八、指令、偏好与适配
0 / 9
九、推理与计算预算
0 / 8
十、检索与上下文
0 / 5
十一、工具与 Agent
0 / 4
十二、生成与高效服务
0 / 5
十三、多模态模型
0 / 6
十四、可信性与机制解释
0 / 5
2026











