def forward(
self,
input: rlt.PreprocessedRankingInput,
mode: str,
tgt_seq_len: Optional[int] = None,
greedy: Optional[bool] = None,
):
res = self.data_parallel(
input, mode=mode, tgt_seq_len=tgt_seq_len, greedy=greedy
)
if mode == Seq2SlateMode.RANK_MODE:
return rlt.RankingOutput(
ranked_per_symbol_probs=res[0],
ranked_per_seq_probs=res[1],
ranked_tgt_out_idx=res[2],
)
elif mode in (
Seq2SlateMode.PER_SYMBOL_LOG_PROB_DIST_MODE,
Seq2SlateMode.PER_SEQ_LOG_PROB_MODE,
):
return rlt.RankingOutput(log_probs=res)
elif mode == Seq2SlateMode.ENCODER_SCORE_MODE:
return rlt.RankingOutput(encoder_scores=res)
else:
raise NotImplementedError()
def forward(
self,
input: rlt.PreprocessedRankingInput,
mode: Seq2SlateMode,
tgt_seq_len: Optional[int] = None,
greedy: Optional[bool] = None,
):
if mode == Seq2SlateMode.RANK_MODE:
# pyre-fixme[16]: `Seq2SlateNet` has no attribute `seq2slate`.
res = self.seq2slate(
mode=mode.value,
state=input.state.float_features,
src_seq=input.src_seq.float_features,
tgt_seq_len=tgt_seq_len,
greedy=greedy,
)
return rlt.RankingOutput(
ranked_per_symbol_probs=res.ranked_per_symbol_probs,
ranked_per_seq_probs=res.ranked_per_seq_probs,
ranked_tgt_out_idx=res.ranked_tgt_out_idx,
)
elif mode in (
Seq2SlateMode.PER_SYMBOL_LOG_PROB_DIST_MODE,
Seq2SlateMode.PER_SEQ_LOG_PROB_MODE,
):
assert input.tgt_in_seq is not None
assert input.tgt_in_idx is not None
assert input.tgt_out_idx is not None
res = self.seq2slate(
mode=mode.value,
state=input.state.float_features,
src_seq=input.src_seq.float_features,
# pyre-fixme[16]: `Optional` has no attribute `float_features`.
tgt_in_seq=input.tgt_in_seq.float_features,
tgt_in_idx=input.tgt_in_idx,
tgt_out_idx=input.tgt_out_idx,
)
if res.per_symbol_log_probs is not None:
log_probs = res.per_symbol_log_probs
else:
log_probs = res.per_seq_log_probs
return rlt.RankingOutput(log_probs=log_probs)
elif mode == Seq2SlateMode.ENCODER_SCORE_MODE:
assert input.tgt_out_idx is not None
res = self.seq2slate(
mode=mode.value,
state=input.state.float_features,
src_seq=input.src_seq.float_features,
tgt_out_idx=input.tgt_out_idx,
)
return rlt.RankingOutput(encoder_scores=res.encoder_scores)
else:
raise NotImplementedError()
def forward(
self,
input: rlt.PreprocessedRankingInput,
mode: str,
greedy: Optional[bool] = None,
):
# The creation of evaluation data pages only uses these specific arguments
assert mode in (Seq2SlateMode.RANK_MODE, Seq2SlateMode.PER_SEQ_LOG_PROB_MODE)
if mode == Seq2SlateMode.RANK_MODE:
assert greedy
return rlt.RankingOutput(
ranked_tgt_out_idx=torch.tensor([[2, 3], [3, 2], [2, 3]]).long()
)
return rlt.RankingOutput(log_probs=torch.log(torch.tensor([0.4, 0.3, 0.7])))