{"id":"arxiv:2401.17401","title":"Step-size Optimization for Continual Learning","date":"2024-01-30","section":"Data, training, optimization","quality":-0.669005,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c114ace9d4e8c21e4f5e7bf6fdaf29c0df7b9527d2935c401329991ca92e94fd/","source_url":"https://arxiv.org/abs/2401.17401","text_available":true,"authors":["Thomas Degris","Khurram Javed","Arsalan Sharifnassab","Yuxin Liu"],"outgoing":[],"incoming":[{"source":"arxiv:2402.02342","target":"arxiv:2401.17401","evidence":"id","id":"arxiv:2402.02342","title":"MetaOptimize: A Framework for Optimizing Step Sizes and Other Meta-parameters","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/6801d4fe20b85924aed2dec1855a35232b2ec1d98e15ec3bc27a0fbe5dfd1af7/"}],"text_parts":[{"start":0,"end":15889,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c114ace9d4e8c21e4f5e7bf6fdaf29c0df7b9527d2935c401329991ca92e94fd/part-1.html"},{"start":15889,"end":26583,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c114ace9d4e8c21e4f5e7bf6fdaf29c0df7b9527d2935c401329991ca92e94fd/part-2.html"}],"contents":[{"title":"Step-size Optimization for Continual Learning","offset":0,"level":1,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c114ace9d4e8c21e4f5e7bf6fdaf29c0df7b9527d2935c401329991ca92e94fd/part-1.html#text"},{"title":"Abstract","offset":49,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c114ace9d4e8c21e4f5e7bf6fdaf29c0df7b9527d2935c401329991ca92e94fd/part-1.html#text"},{"title":"1 The Role of Step-size in Continual Learning","offset":1241,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c114ace9d4e8c21e4f5e7bf6fdaf29c0df7b9527d2935c401329991ca92e94fd/part-1.html#text"},{"title":"2 Setting","offset":7124,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c114ace9d4e8c21e4f5e7bf6fdaf29c0df7b9527d2935c401329991ca92e94fd/part-1.html#text"},{"title":"3 Limitations of Step-size Normalization","offset":10612,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c114ace9d4e8c21e4f5e7bf6fdaf29c0df7b9527d2935c401329991ca92e94fd/part-1.html#text"},{"title":"4 Step-size Optimization with IDBD","offset":16317,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c114ace9d4e8c21e4f5e7bf6fdaf29c0df7b9527d2935c401329991ca92e94fd/part-2.html#text"},{"title":"5 Limitations of IDBD","offset":20251,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c114ace9d4e8c21e4f5e7bf6fdaf29c0df7b9527d2935c401329991ca92e94fd/part-2.html#text"},{"title":"6 Existing Extensions of IDBD","offset":21231,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c114ace9d4e8c21e4f5e7bf6fdaf29c0df7b9527d2935c401329991ca92e94fd/part-2.html#text"},{"title":"7 Other Step-size Adaptation Algorithms","offset":22602,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c114ace9d4e8c21e4f5e7bf6fdaf29c0df7b9527d2935c401329991ca92e94fd/part-2.html#text"},{"title":"8 A Promising Research Direction: Normalized Step-size Optimization","offset":25189,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c114ace9d4e8c21e4f5e7bf6fdaf29c0df7b9527d2935c401329991ca92e94fd/part-2.html#text"}],"text_code_points":26583,"source_markdown":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c114ace9d4e8c21e4f5e7bf6fdaf29c0df7b9527d2935c401329991ca92e94fd/source.md"}