💫 Patch v0.0.3

Release Note (0.0.3)

Release time: 2021-10-19 23:04:12

🙇 We'd like to thank all contributors for this new release! In particular,
Han Xiao, Wang Bo, Maximilian Werk, Tadej Svetina, Alex Cureton-Griffiths, Roshan Jossy, Jina Dev Bot, 🙇

🆕 New Features

• [84585bee] - refactor head layers (#130) (Tadej Svetina)
• [b624a62a] - tuner: allow adjustment of optimizer (#128) (Tadej Svetina)
• [98c584e4] - enable saving of models in backend (#115) (Maximilian Werk)
• [2296ca09] - tuner: add gpu for paddle and tf (#121) (Wang Bo)
• [c60ec838] - tuner: add gpu support for pytorch (#122) (Tadej Svetina)
• [c971f824] - logging of train and eval better aligned (#105) (Maximilian Werk)
• [a6d16ff2] - tailor: add display and refactor summary (#112) (Han Xiao)
• [bd4cfff4] - fit: add tailor to top-level fit function (#108) (Han Xiao)
• [82c2cc8d] - tailor: attach a dense layer to tailor (#96) (Wang Bo)
• [04de292a] - tailor: add high-level framework-agnostic convert (#97) (Han Xiao)

⚡ Performance Improvements

• [68fc7839] - tuner: inference mode for torch evaluation (#89) (Tadej Svetina)

🐞 Bug fixes

• [b951426d] - change helper function to private (#151) (Han Xiao)
• [bc8b36ef] - demo: fix celeba example docs, logic, code (#145) (Han Xiao)
• [ed6d8c67] - frontend layout tweaks (#142) (Han Xiao)
• [02852803] - overfit test (#137) (Tadej Svetina)
• [5a25a729] - helper: add real progressbar for training (#136) (Han Xiao)
• [5196ce2a] - api: add kwargs to fit (#95) (Han Xiao)
• [1a8272ca] - threading also for gateway (#83) (Maximilian Werk)
• [e170d95b] - cd: fix prerelease script (Han Xiao)

🧼 Code Refactoring

• [2916e9f5] - tuner: revert some catalog change before release (#150) (Han Xiao)
• [635cd4c2] - adjust type hints (#149) (Wang Bo)
• [b67ab1a5] - helper: move get_tailor and get_tunner to helper (#134) (Han Xiao)
• [052adbb2] - helper: move get_tailor and get_tunner to helper (#131) (Han Xiao)
• [d8ff3a5b] - labeler UI: js file into components (#101) (Roshan Jossy)
• [80b5a2a1] - tailor: rename convert function to_embedding_model (#103) (Han Xiao)
• [c06292cb] - tailor: use different trim logic (#100) (Han Xiao)
• [1956a3d3] - tailor: fix type hint in tailor (#88) (Han Xiao)
• [91587d88] - tailor: improve interface (#82) (Han Xiao)
• [56eb5e8f] - api: move fit into top-most init (#84) (Han Xiao)

📗 Documentation

• [c2584876] - add catalog to docs (#147) (Maximilian Werk)
• [6fd3e1ea] - tuner: add docstrings (#148) (Tadej Svetina)
• [177a78dd] - fix generate docs (#144) (Maximilian Werk)
• [ac2d23de] - polish (#146) (Alex Cureton-Griffiths)
• [b0da1bf6] - add celeba example (#143) (Wang Bo)
• [475c1d8b] - tuner: add loss function explain for tuner (#138) (Han Xiao)
• [f47e27a3] - update banner hide design (Han Xiao)
• [11a6a8b9] - add interactive selector (Han Xiao)
• [08ba5e06] - add tailor feature image (Han Xiao)
• [528c80d5] - tailor: add docs for tailor (#119) (Han Xiao)
• [04c22f74] - tailor: add first draft on tailor (Han Xiao)
• [e62f77ea] - helper: add docstring for types (#98) (Han Xiao)

🏁 Unit Test and CICD

• [6b8eca8c] - use jina git source as test dependencies (#135) (Han Xiao)
• [f91f39f5] - add tailor plus tuner integration test (#124) (Wang Bo)
• [56c13e59] - add pr labeler (#123) (Han Xiao)
• [562c65f5] - tuner: add test for overfitting (#109) (Tadej Svetina)
• [b448a611] - tailor: assure weights are preserved after calling to_embedding_model (#106) (Wang Bo)
• [47b7a55d] - tailor: add test for name is none (#87) (Wang Bo)

🍹 Other Improvements

• [370e5fba] - cd: add tag and release note script (Han Xiao)
• [33b1c90b] - update readme (Han Xiao)
• [0be69a45] - Introduce catalog + ndcg (#120) (Maximilian Werk)
• [8bba726e] - update svg (Han Xiao)
• [dfc334f7] - fix emoji (Han Xiao)
• [a589a016] - docs: add note from get_framework (Han Xiao)
• [d970a2b6] - fix styling (Han Xiao)
• [62a0da7e] - version: the next version will be 0.0.3 (Jina Dev Bot)