MCPcopy Create free account
hub / github.com/PaddlePaddle/FastDeploy / ModelOutputData

Class ModelOutputData

fastdeploy/worker/output.py:199–353  ·  view source on GitHub ↗

OutputData by execute_model

Source from the content-addressed store, hash-verified

197
198@dataclass
199class ModelOutputData:
200 """
201 OutputData by execute_model
202 """
203
204 """
205 Tokens generated in the previous step
206 """
207 next_tokens: paddle.Tensor
208
209 """
210 Flags indicating whether decoding should stop
211 """
212 stop_flags: paddle.Tensor
213
214 """
215 Index of the current decoding step
216 """
217 step_idx: int
218
219 """
220 Maximum decoding length
221 """
222 max_dec_len: int
223
224 """
225 Previous ids used for decoding
226 """
227 pre_ids: paddle.Tensor
228
229 """
230 Sequence lengths for this step
231 """
232 seq_lens_this_time: paddle.Tensor
233
234 """
235 Eos token ID
236 """
237 eos_token_id: paddle.Tensor
238
239 """
240 Indicates if stopping conditions should be ignored
241 """
242 not_need_stop: bool
243
244 """
245 Sequence lengths of the encoder
246 """
247 seq_lens_encoder: paddle.Tensor
248
249 """
250 Sequence lengths of the decoder
251 """
252 seq_lens_decoder: paddle.Tensor
253
254 """
255 Indicates if this is a blocking step
256 """

Callers 11

_dummy_runMethod · 0.90
execute_modelMethod · 0.90
_dummy_runMethod · 0.90
execute_modelMethod · 0.90
_dummy_pooler_runMethod · 0.90
_dummy_sampler_runMethod · 0.90
_postprocessMethod · 0.90
execute_modelMethod · 0.90
_dummy_pooler_runMethod · 0.90
_dummy_sampler_runMethod · 0.90
_postprocessMethod · 0.90

Calls

no outgoing calls

Tested by

no test coverage detected