QuestionAnsweringTrainer cannot produce answers with characters above U+00FF (answer is decoded one byte per character)

オープン
  • 何か問題がありますか？
#7,749 コメント 1 件 リアクション 0 件 担当者 0 名 GitHub で見る

メンテナーはふだん 1 日以内に返信

まだ誰も着手していません。

評価

難易度
3/5
見積もり時間
1〜2日
初心者へのやさしさ
74/100
issue の種類
バグ
明瞭さ
明確に書かれている
活発さ
活発
技術スタック
csharp
領域
machine-learning

調査の方向性

Start in Roberta/QATrainer.cs, where the mapper fills PredictedAnswersBuffer, and reproduce the issue with the Korean SQuAD-format dataset described in the report. Inspect the tokenizer offsets from EncodeToTokens and the TokenizerExtensions.GetInstance path. Done means predicted answers preserve non-Latin characters and match the corresponding context substrings, with the existing reproduction completing normally.

索引モデルが issue の本文から書いたものです。

説明

area-TorchSharp untriaged
Description

QuestionAnsweringTrainer (Microsoft.ML.TorchSharp, Roberta/QATrainer.cs) cannot produce an answer that contains characters above U+00FF. For Korean, Chinese, Japanese, Cyrillic and similar text, the predicted answer comes out with those characters missing, so exact match is unreachable whatever the model learned.

Cause

The predicted answer is not sliced out of the context string. It is decoded from the span's token ids:

// QATrainer.cs (main), in the mapper that fills PredictedAnswersBuffer
_parent.Tokenizer.Decode(
    _parent.Tokenizer.RobertaModel().ConvertOccurrenceRanksToIds(contextIds).ToArray()
        .AsSpan(predictStart - questionLength - 2, predictEnd - predictStart).ToArray())

The tokenizer comes from TokenizerExtensions.GetInstance, which calls EnglishRobertaTokenizer.Create(...) with the default filterUnsupportedChars = true. On that path EnglishRobertaTokenizer.MapIdToToken maps each token character through UnicodeToByte and appends the byte value as a char. Every output character is therefore ≤ U+00FF, and multi-byte UTF-8 sequences are never reassembled into the original characters.

Reproduction
  1. Train mlContext.MulticlassClassification.Trainers.QuestionAnswer(...) on a small SQuAD-format dataset whose contexts and answers are in Korean (any non-Latin script shows the same behavior).
  2. Transform a test set and read the Answer column.

Observed with 26 training rows and 10 epochs: training completes normally, but all 8 predicted answers contain only the digits, ASCII punctuation and spaces of their spans, with every Hangul character missing, e.g. "12 , 1 100". Character-level F1 is about 0.05–0.08 and exact match is 0.

Expected behavior

The predicted answer is the corresponding substring of the context, in whatever script the context uses.

Suggested fix

Map the predicted token span back to character offsets in the original context (the tokenizer already provides offsets via EncodeToTokens) and return context.Substring(start, length), instead of decoding the ids. Alternatively, decode byte-level tokens to bytes and UTF-8-decode them (filterUnsupportedChars: false with proper byte reassembly).

Environment
  • Microsoft.ML 5.0.0, Microsoft.ML.TorchSharp 0.23.0 (behavior also matches the current main source)
  • .NET 10, Windows 11, CPU
主要言語
C#
スター
9.4k
フォーク
2k
平均マージ
2日 18時間
マージ済み PR（30日）
64

環境構築

はじめの一歩

  1. issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
  2. 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
  3. リポジトリをフォークし、ブランチを切って変更します。
  4. issue 番号を参照したプルリクエストを送ります。

dotnet/machinelearning のほかの issue

dotnet/machinelearning の issue をすべて見る

似ている issue

C# の issue をもっと見る

新しい issue をメールで受け取る

初心者向けの GitHub issue を短くまとめたダイジェスト。