Ikibazo gihuza ibibazo · Biragoye Urwego
Ikibazo cya XLNet
Kugenzura impushya zo kwemeza, ibyerekezo-bibiri byitondewe, Transformer-XL isubiramo, nuburyo ibipimo byerekana biterwa nigenamiterere rigereranijwe.
Ikibazo 1 Bya 8
Niyihe ntego yo kwitegura XLNet ikoresha?
Reba buri kibazo8 ibibazo
Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.
- Niyihe ntego yo kwitegura XLNet ikoresha?
- Ururimi rwerekana uruhushya
- Kwerekana imiterere y'ururimi
- Gusimbuza ibimenyetso
- Guhanura interuro ikurikira
- Ni izihe ntege nke za BERT XLNet yagenewe gukemura?
- BERT yari nto cyane
- Ikimenyetso [MASK] gitera kwitiranya / guhuza neza no gufata ubwigenge bwa maskike-token
- BERT ntabwo yashoboye gukora interuro ngufi
- BERT yabuze kodegisi
- Ni ubuhe butumwa bwa XLNet bwo-kwiyitaho butandukanye?
- Encoder na decoder
- Amahugurwa n'ibishushanyo mbonera
- Umugezi wibirimo hamwe numurongo wibibazo
- Gushyiramo interuro ninteruro
- Ese XLNet ihindura umubiri mumagambo?
- Nibyo, biratondekanya ibimenyetso bifatika
- Nibyo, ariko mugihe cyo gutunganya neza
- Oya, ntabwo ihindura ibyateganijwe na gato
- Oya, yemerera gutondekanya ibintu ukoresheje mask yo kwitondera mugihe ukomeje imyanya
- Ni ubuhe bwubatsi bukora nkumugongo wa XLNet kumurongo muremure?
- Guhindura-XL
- LSTM
- Umwimerere wa vanilla Transformer
- Umuyoboro uhuza
- How does XLNet combine bidirectional context with autoregressive prediction?
- It predicts all tokens independently with no attention mask
- It predicts in sampled factorization orders without inserting a [MASK] input token, using attention masks
- It physically shuffles the sentence before every prediction
- It replaces autoregressive prediction with a denoising autoencoder
- For a token sequence A B C, suppose a sampled factorization order predicts C before A and B. What context is available when predicting C?
- The tokens A and B must always be provided first
- Only the next sentence is available
- No earlier tokens in that sampled order, although C remains at its original input position
- The input sequence is physically reordered to C A B
- How does XLNet prevent the target token from leaking into its own prediction query?
- It removes the target from the entire training sequence
- It uses the target’s label as an input embedding
- It replaces the token with a [MASK] symbol
- The query stream represents the target position without its token content