Based on the scenario, which single parameter within the logarithmic bucketing formula is the most probable cause of this issue, and why?

Google

In the context of machine translation, the auto-regressive decoding process generates each token of the translated text (target language) by conditioning on both the complete source language sequence and the tokens that have already been generated in the target language.

Auto-regressive Decoding in Machine Translation

A sequence-to-sequence model is translating a sentence from one language to another. At the third step of generating the output, it incorrectly predicts a word. Subsequent words generated by the model are also incorrect, leading to a final translation that is grammatically flawed and semantically nonsensical, even though the initial error was minor. Which characteristic of the generation process is the most likely cause of this cascading failure?

Diagnosing Long-Range Dependency Issues

A key characteristic of the text generation process in many machine translation systems is that each new word is produced based on the words that came before it. Analyze the primary trade-off this sequential, step-by-step approach introduces, specifically contrasting its impact on the computational demands of the process with its effect on the coherence of the final translated output.

Learn Before

Related