Skip to content

TubeTran v0.1.16

Choose a tag to compare

@github-actions github-actions released this 17 Nov 10:33
· 13 commits to main since this release

本版重點 / Highlights

  • 修正:按全螢幕鈕後,焦點會回到播放器。 以前焦點停在全螢幕鈕上,← → 無法倒轉、快轉;現在進入全螢幕就能直接用(投影片模式則是上一則、下一則)。

  • 改善:翻譯失敗時看得到原因,也能直接重試。 錯誤訊息會附上最後一次失敗的原因(例如「回覆超過上限被截斷」「回覆少了幾句」);狀態列下方有「重試」鈕。影片庫的卡片會標成「翻譯失敗」並有重試鈕,滑鼠移上去可看到原因。重試從中斷處接著翻,已翻好的句子不會重翻。

  • 修正:降低翻譯失敗的機會。 每批的輸出上限從 8192 提高到 16000 個 token;模型會先花一部分 token 思考,原本的上限偶爾不夠,譯文被截斷就會三次都失敗。

  • Fixed: focus returns to the player after pressing the fullscreen button. Previously focus stayed on the button, so ← → didn't seek; now they work right away in fullscreen (in slide mode they go to the previous / next line).

  • Improved: failed translations show the reason and can be retried. The error now includes why the last attempt failed (for example "the reply was cut off at the limit" or "the reply was missing lines"), with a "Retry" button below the status line. The library card is marked "Translation failed" with a retry button, and hovering shows the reason. Retrying continues where it stopped; lines already translated are kept.

  • Fixed: translation fails less often. The per-batch output limit was raised from 8192 to 16000 tokens; the model spends part of it thinking first, and the old limit was occasionally too small, cutting off the reply so all three attempts failed.

Full Changelog: v0.1.15...v0.1.16