Mit dem Guide verknüpftes Quiz · Mittel Ebene
Jasper und QuartzNet ASR-Quiz
Vergleichen Sie die Faltungsdesigns und gemeldeten Konfigurationen von Jasper und QuartzNet und testen Sie dann die CTC-Dekodierung und die Arithmetik trennbarer Filter.
Frage 1 von 8
Which factorization distinguishes QuartzNet’s time-channel separable convolution from a dense temporal convolution?
Sehen Sie sich jede Frage in der Vorschau an8 Fragen
Während Sie am Quiz teilnehmen, werden Antworten und Erklärungen angezeigt.
- Which factorization distinguishes QuartzNet’s time-channel separable convolution from a dense temporal convolution?
- Replacing every neural layer with a GMM
- Removing all channels from the input
- Adding a mandatory recurrent connection at every step
- Separate depthwise temporal filtering and pointwise channel mixing
- In the QuartzNet paper’s stated comparison, which approximate count is listed for QuartzNet-15x5 versus JasperDR-10x5 at 333 million parameters?
- 250 million
- 500 Millionen
- 19 million
- One billion
- Which organization is associated with the cited Jasper and QuartzNet research papers?
- NVIDIA
- Amazon
- OpenAI
- Welche Verlustfunktion verwenden Jasper und QuartzNet zum Trainieren ohne explizite Ausrichtungen?
- Konnektionistische Zeitklassifikation (CTC)
- Mittlerer quadratischer Fehler
- Kreuzentropie auf festen Rahmen
- Triplettverlust
- Which structural feature is described for the Jasper convolutional stack?
- Only alphabetical sorting of frames
- A guarantee that all gradients have equal magnitude
- Residual connections
- Random deletion of every deep layer
- Which augmentation combination does the QuartzNet paper report for its best evaluated training setup?
- Only deleting all training transcriptions
- A required pronunciation dictionary replacement at each frame
- A guarantee that no data augmentation is used
- Speed perturbation with a cutout-style spectrogram method
- Under the stated CTC path mapping, what labels result from A, A, blank, A?
- AA
- A
- AAA
- An empty sequence
- For a length-three depthwise temporal filter on four input channels plus a pointwise map to eight output channels, how many filter weights are counted, excluding biases?
- 96
- 44
- 12
- 32