论文阅读
文章平均质量分 93
Isawany
默默读书默默长大
展开
-
论文笔记--Llama3 report
Meta最新开源模型Llama3-8B, Llama3-70B原创 2024-04-22 11:40:33 · 911 阅读 · 1 评论 -
论文笔记--Learning Political Polarization on Social Media Using Neural Networks
IOM-NN:更准确的极化预测方法原创 2023-12-23 15:28:47 · 1332 阅读 · 0 评论 -
论文笔记--Gemini: A Family of Highly Capable Multimodal Models
Gemini-现存最强大的多模态模型原创 2023-12-08 17:17:33 · 1535 阅读 · 0 评论 -
论文笔记--InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning
InstructBLIP-基于指令微调的BLIP系列多模态模型原创 2023-12-07 16:33:29 · 996 阅读 · 0 评论 -
论文笔记--A Fine-grained Interpretability Evaluation Benchmark for Neural NLP
中、英文的NLP benchmark原创 2023-12-06 16:17:36 · 1251 阅读 · 0 评论 -
论文笔记--Increasing Diversity While Maintaining Accuracy: Text Data Generation with Large Language Mode
标题:Increasing Diversity While Maintaining Accuracy: Text Data Generation with Large Language Models and Human Interventions作者:John Joon Young Chung, Ece Kamar, Saleema Amershi日期:2023。原创 2023-11-30 16:47:44 · 1062 阅读 · 0 评论 -
论文笔记--Exploiting Asymmetry for Synthetic Training Data Generation: SynthIE and the Case of Informati
自动生成知识抽取的数据,得到更加干净、均匀的数据集原创 2023-11-29 11:31:12 · 997 阅读 · 0 评论 -
论文笔记--Fly-Swat or Cannon? Cost-Effective Language Model Choice via Meta-Modeling
CELMOC-自动为你的query分配合适的LM,在不降低性能的前提下减轻推理cost原创 2023-11-28 16:13:54 · 940 阅读 · 0 评论 -
论文笔记--Toolformer: Language Models Can Teach Themselves to Use Tools
Toolformer-一个可自动访问API的语言模型工具原创 2023-11-26 19:50:35 · 1244 阅读 · 0 评论 -
论文笔记--DetectGPT: Zero-Shot Machine-Generated Text Detection using Probability Curvature
DetectGPT-通过随机扰动检测文本是否为机器生成原创 2023-11-25 19:22:49 · 1195 阅读 · 0 评论 -
论文笔记--ERNIE-M: Enhanced Multilingual Representation by Aligning Cross-lingual Semantics with Monolin
ERNIE-M:通过跨注意力机制对齐不同语言的信息原创 2023-11-14 19:30:41 · 77 阅读 · 0 评论 -
论文笔记--Baichuan 2: Open Large-scale Language Models
百川2-领先的开源大模型原创 2023-11-12 11:23:18 · 781 阅读 · 0 评论 -
Joint Dropout: Improving Generalizability in Low-Resource Neural Machine Translation through Phrase
JD-提高低资源语言的NMT的鲁棒性原创 2023-10-19 08:21:48 · 88 阅读 · 0 评论 -
论文笔记--Enriching Word Vectors with Subword Information
FastText,将subword信息融入skipgram模型原创 2023-10-09 07:54:59 · 757 阅读 · 1 评论 -
论文笔记--Augmenting Pre-trained Language Models with QA-Memory for Open-Domain Question Answering
QAMAT:基于两阶段训练的QA模型原创 2023-08-20 20:36:12 · 184 阅读 · 0 评论 -
论文笔记--Llama 2: Open Foundation and Fine-Tuned Chat Models
llama2最全最细剖析:开源LLM更上一层楼原创 2023-08-12 22:51:02 · 1395 阅读 · 2 评论 -
论文笔记--ERNIE-VIL 2.0: MULTI-VIEW CONTRASTIVE LEARNING FOR IMAGE-TEXT PRE-TRAINING
ERNIE-ViL 2.0:基于多视角学习的多模态学习原创 2023-07-31 20:05:16 · 387 阅读 · 0 评论 -
论文笔记--ERNIE-ViL: Knowledge Enhanced Vision-Language Representations through Scene Graphs
ERNIE-ViL:首次引入Scene graph训练的多模态模型原创 2023-07-30 16:48:25 · 261 阅读 · 0 评论 -
论文笔记--GloVe: Global Vectors for Word Representation
GloVe:一种结合统计量和上下文窗口的词嵌入方法原创 2023-07-27 21:38:36 · 1534 阅读 · 0 评论 -
论文笔记--FEDERATED LEARNING: STRATEGIES FOR IMPROVING COMMUNICATION EFFICIENCY
基于预置结构和模型压缩的联邦学习,实现客户端和中央服务器的更高效的交流原创 2023-07-27 19:56:47 · 1497 阅读 · 0 评论 -
论文笔记--Skip-Thought Vectors
Skip-thought:一种基于Encoder-decoder的句向量预训练模型原创 2023-07-25 22:17:23 · 1187 阅读 · 0 评论 -
论文笔记--Distilling the Knowledge in a Neural Network
distill:模型压缩的新范式原创 2023-07-23 11:11:02 · 1413 阅读 · 0 评论 -
论文笔记--ERNIE: Enhanced Language Representation with Informative Entities
ERNIE:将知识图谱融合进预训练的模型原创 2023-07-21 11:02:27 · 588 阅读 · 0 评论 -
论文笔记--Won’t Get Fooled Again: Answering Questions with False Premises
FalseQA:第一份人造的伪命题问答数据集原创 2023-07-20 23:28:56 · 605 阅读 · 0 评论 -
论文笔记--Knowledgeable Prompt-tuning: Incorporating Knowledge into Prompt Verbalizer for Text Classific
KPT:一种有效融合外部知识的prompt tuning手段原创 2023-07-20 15:18:27 · 155 阅读 · 0 评论 -
论文笔记--OpenPrompt: An Open-source Framework for Prompt-learning
OpenPrompt:一款开源prompt learning工具原创 2023-07-18 20:51:35 · 1388 阅读 · 0 评论 -
论文笔记--PTR: Prompt Tuning with Rules for Text Classification
PTR:利用子任务sub-prompts的聚合得到高效的prompt原创 2023-07-16 14:38:28 · 652 阅读 · 0 评论 -
论文笔记--SentencePiece: A simple and language independent subword tokenizer and detokenizer for Neural
SentencePiece原文解析和Python代码实现原创 2023-07-13 21:56:09 · 204 阅读 · 0 评论 -
论文笔记--TinyBERT: Distilling BERT for Natural Language Understanding
TinyBERT:基于Transformer蒸馏两阶段BERT蒸馏模型原创 2023-07-12 22:22:17 · 820 阅读 · 0 评论 -
论文笔记--DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter
DistilBERT:40%的参数实现对BERT的蒸馏原创 2023-07-10 22:17:27 · 206 阅读 · 1 评论 -
论文笔记--SentEval: An Evaluation Toolkit for Universal Sentence Representations
SentEval:自动评估NLP任务的性能,实现NLP学术研究成果评估的一致性原创 2023-07-09 17:35:20 · 1163 阅读 · 0 评论 -
论文笔记--Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
T5模型:18/24个NLP任务上达到SOTA,对大模型的影响因子进行系统论证原创 2023-07-08 15:57:02 · 190 阅读 · 0 评论 -
论文笔记--Goat: Fine-tuned LLaMA Outperforms GPT-4 on Arithmetic Tasks
Goat:一种可以高精度进行数学运算的大语言模型原创 2023-07-01 22:58:35 · 2408 阅读 · 2 评论 -
论文笔记--kNN PROMPTING: BEYOND-CONTEXT LEARNING WITH CALIBRATION-FREE NEAREST NEIGHBOR INFERENCE
kNN prompting:利用全部标记数据进行推理原创 2023-06-28 23:19:53 · 543 阅读 · 0 评论 -
论文笔记--On the Sentence Embeddings from Pre-trained Language Models
BERT系列模型阅读之BERT-flow:一种解决BERT嵌入空间各向异性的变换方法原创 2023-06-24 23:39:19 · 487 阅读 · 0 评论 -
论文笔记--Prompt Consistency for Zero-Shot Task Generalization
Swarm Distillation: 基于prompt一致性的无监督学习原创 2023-06-23 18:47:45 · 1146 阅读 · 1 评论 -
论文笔记--STRUCTBERT: INCORPORATING LANGUAGE STRUCTURES INTO PRE-TRAINING FOR DEEP LANGUAGE UNDERSTANDIN
BERT系列文章阅读之StructBERT:以语言结构为学习任务的BERT系列模型原创 2023-06-19 22:17:42 · 123 阅读 · 1 评论 -
论文笔记--LIMA: Less Is More for Alignment
LIMA:对齐样本不是越多越好,小样本也可以超越数据堆积原创 2023-06-13 21:03:04 · 1428 阅读 · 1 评论 -
论文笔记--GPT-4 Technical Report
GPT-4报告:更强大的多模态模型原创 2023-06-11 17:25:28 · 3122 阅读 · 3 评论 -
论文笔记--SimCSE: Simple Contrastive Learning of Sentence Embeddings
通过dropout进行对比学习:SimCSE最全公式推导原创 2023-06-10 21:35:11 · 909 阅读 · 1 评论