NanaEilish727

@NanaEilish727

Joined on Sep 7, 2022

  • 好像不完全算 paper reading,算 theory or algo-understanding。 以下都是一堆數學,簡單來說就是一堆機率統計(包含貝氏)和以此延伸的最佳化。 🎁 Introduction Notations Two Assumptions held for the first order HMM Markov Assumption: $P(q_i|q_1, ..., q_{i-1}) = P(q_i|q_{i-1})$ 轉換到現在 state,只與前一個 state 有關。 Output Independence: $P(o_i|q_{1:T}, o_{1:T}) = P(o_i|q_i)$ output observation $o_i$ 只和當前的 state $q_i$(state that produced $o_i$) 有關,和其他任何 states 或任何 observations 都無關。
     Like  Bookmark
  • 如何使用 demo.sh 產出結果並呈現於前端 project_root = RobertaABSA Downloads: 請先下載並放在對應的位置。 Drive Link For tree-induction: RobertaABSA/Train/save_models/roberta-en-Laptop-FT/roberta-en RobertaABSA/Train/save_models/bert-en-base-uncased-Laptop-FT/bert-en-base-uncased For aspect-level sentiment prediction: RobertaABSA/ASGCN/state_dict
     Like  Bookmark
  • Ruotian Ma et al. 上海復旦大學 2022 Naacl https://github.com/rtmaww/EntLM/ 0. How's it related to our project ChemiCloud (1) few-shot setting:資料量少的 supervised learning, 化學雲官方可能無力標注大量資料。或甚至需要我們自己標(一百句左右)。如果懶得架設標記平台或大量正式標記不可得的話,few-shot 的設定基礎很重要。 (2) 似乎是首篇論文研究 Prompt-tuning/Large LMs 在 NER 任務上如何有效率 reformulate 並訓練。 2023.2.2 化學雲報告投影片(有四頁介紹這篇)
     Like  Bookmark
  • Wenxuan Zhang et al. (Hong Kong University) DAMO Academy, Alibaba Group Paper: https://aclanthology.org/2021.findings-emnlp.390.pdf Code: https://github.com/IsakZhang/ABSA-QA 1. Introduction ABSA = ATE + ASC (same task-splitting as our current system) ABSA-QA (the ABSA task on QA forums, where QA forum means an opinion-sharing platform have comment section that allows people to ask questions and answer) 2. Methodology
     Like  Bookmark
  • Papers 經典:📃 DDPM paper (Jonathan Ho et al, 2020) NCKU 資工所深度學習課程報告選讀論文:📃 improved DDPM (Alex Nichol et al, 2021) Notes 💡學長的投影片 塗塗 Other References 邊實作邊學習diffusion-model-從ddpm的簡化概念理解
     Like  Bookmark
  • Wei, Jason et al. from Google Research Published as a conference paper at ICLR Github: https://github.com/google-research/flan. Arxiv: https://arxiv.org/abs/2109.01652 Intro The motivation of instruction tuning is to improve the ability of language models to respond to NLPinstructions. Transform existing datasets into instruction format.
     Like  Bookmark
  • P76114511 楊晴雯 Testing environment: MacOSX with Intel Core, VSCode, Python 3.8 I. Analysis Data Source: the 2022 released testdata 2,997 transactions 50 items in total 5 items in 1 transaction in average
     Like  Bookmark
  • If you have any questions, please e-mail to nckudm@gmail.com. Topic: Classification Analysis Data (35 pts) You should design your own data; using data online will result in 0 points in this section. Problem Definition / Data Design (10 pts) 10 Features (10 pts), you can add redundant features and observe the performance. 5 Rules (15 pts)
     Like  Bookmark
  • P76114511 資訊所一 楊晴雯 Topic: Classification Analysis ⛓ Link to Repo ⛓ Link to HackMD I. 資料設計(Data Design) 這是一個動漫資料集之設計,每一筆資料代表一部動漫作品。給定一組離散(類別或數值)或連續數值的屬性資料,該任務為一三分類任務,三個分類為:鉅作(masterpiece)、小品(refreshing)、平淡(bland),欲判斷「該部動漫為三分類的哪一類」。 (I) 特徵/屬性(Features/Attributes) 以如下11個屬性輸入作為判斷。
     Like 1 Bookmark
  • If you have any questions, please e-mail to nckudm@gmail.com. Topic: Classification Analysis Data (35 pts) You should design your own data; using data online will result in 0 points in this section. Problem Definition / Data Design (10 pts) 10 Features (10 pts), you can add redundant features and observe the performance. 5 Rules (15 pts)
     Like  Bookmark