Skip to content

fix unimo decoding_lib path config #6690

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
huismiling wants to merge 2 commits into from
Closed

fix unimo decoding_lib path config #6690

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
590a77c
fix invalid user commit
huismiling Aug 28, 2023
def2815
Merge branch 'PaddlePaddle:develop' into develop
huismiling Aug 28, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions paddlenlp/ops/fast_transformer/src/fusion_unified_decoding_op.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -66,7 +66,7 @@ std::vector<paddle::Tensor> UnifiedDecodingForward(
const int& num_layer,
const int& bos_id,
const int& eos_id,
const int64_t& max_len,
const int& max_len,
const float& beam_search_diversity_rate,
const int& unk_id,
const int& mask_id,
Expand Down Expand Up @@ -251,7 +251,7 @@ std::vector<std::vector<int64_t>> UnifiedDecodingInferShape(
const int& num_layer,
const int& bos_id,
const int& eos_id,
const int64_t& max_len,
const int& max_len,
const float& beam_search_diversity_rate,
const int& unk_id,
const int& mask_id,
Expand Down Expand Up @@ -397,7 +397,7 @@ PD_BUILD_OP(fusion_unified_decoding)
"num_layer: int",
"bos_id: int",
"eos_id: int",
"max_len: int64_t",
"max_len: int",
"beam_search_diversity_rate: float",
"unk_id: int",
"mask_id: int",
Expand Down
7 changes: 4 additions & 3 deletions paddlenlp/transformers/unimo/modeling.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -104,7 +104,7 @@
if input_ids is not None:
num_pad = paddle.sum((input_ids == self.pad_token_id).astype("float32"), axis=-1, keepdim=True)
position_ids = F.relu(
paddle.expand_as(paddle.arange(end=inputs_shape[1], dtype="int64"), inputs_shape) - num_pad
paddle.expand_as(paddle.arange(end=inputs_shape[1], dtype="float32"), inputs_shape) - num_pad
).astype("int64")
else:
logger.warning(
Expand Down Expand Up @@ -462,6 +462,7 @@
def prepare_fast_entry(self, kwargs):
from paddlenlp.ops import FasterMIRO, FasterUNIMOText

decoding_lib = kwargs.get("decoding_lib", None)

Check warning on line 465 in paddlenlp/transformers/unimo/modeling.py

View check run for this annotation

Codecov / codecov/patch

paddlenlp/transformers/unimo/modeling.py#L465 

Added line #L465 was not covered by tests
use_fp16_decoding = kwargs.get("use_fp16_decoding", False)
decode_strategy = kwargs.get("decode_strategy")
if decode_strategy == "sampling" and kwargs.get("top_k") != 0 and kwargs.get("top_p") != 1:
Expand All @@ -480,9 +481,9 @@
)

if getattr(self.encoder, "norm", None) is None:
self._fast_entry = FasterUNIMOText(self, use_fp16_decoding=use_fp16_decoding).forward
self._fast_entry = FasterUNIMOText(self, use_fp16_decoding=use_fp16_decoding, decoding_lib=decoding_lib).forward

Check warning on line 484 in paddlenlp/transformers/unimo/modeling.py

View check run for this annotation

Codecov / codecov/patch

paddlenlp/transformers/unimo/modeling.py#L484 

Added line #L484 was not covered by tests
else:
self._fast_entry = FasterMIRO(self, use_fp16_decoding=use_fp16_decoding).forward
self._fast_entry = FasterMIRO(self, use_fp16_decoding=use_fp16_decoding, decoding_lib=decoding_lib).forward

Check warning on line 486 in paddlenlp/transformers/unimo/modeling.py

View check run for this annotation

Codecov / codecov/patch

paddlenlp/transformers/unimo/modeling.py#L486 

Added line #L486 was not covered by tests
return self._fast_entry

def adjust_logits_during_generation(self, logits):
Expand Down