riaz.somc/transformers
1
0
Fork 0
mirror of https://github.com/huggingface/transformers.git synced 2025-07-31 02:02:21 +06:00
Code Issues Actions 23 Packages Projects Releases Wiki Activity

Update logits_process.py

Browse Source
This commit is contained in:
Jess 2023-08-12 22:24:37 -07:00 committed by GitHub
parent ab75556e28
commit 5f3eb85aa1
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 8 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
src/transformers/generation/logits_process.py
View File

@ -1085,17 +1085,19 @@ class PrefixConstrainedLogitsProcessor(LogitsProcessor):
class HammingDiversityLogitsProcessor(LogitsProcessor):
r"""
[`LogitsProcessor`] that enforces diverse beam search. Note that this logits processor is only effective for
[`LogitsProcessor`] that enforces diverse beam search.
Note that this logits processor is only effective for
[`PreTrainedModel.group_beam_search`]. See [Diverse Beam Search: Decoding Diverse Solutions from Neural Sequence
Models](https://arxiv.org/pdf/1610.02424.pdf) for more details.
<Tip> again: this logits processor is only effective for [`PreTrainedModel.group_beam_search`]. </Tip>
<Tip>
Diverse beam search can be particularly useful in scenarios where a variety of different outputs is desired, rather than multiple similar sequences. It allows the model to explore different generation paths and provides a broader coverage of possible outputs.
Diverse beam search can be particularly useful in scenarios where a variety of different outputs is desired, rather than multiple similar sequences.
It allows the model to explore different generation paths and provides a broader coverage of possible outputs.
</Tip>
<Warning>
This logits processor can be resource-intensive, especially when using large models or long sequences.
@ -1166,7 +1168,7 @@ class HammingDiversityLogitsProcessor(LogitsProcessor):
# Set up for diverse beam search
num_beams = 6
num_beam_groups = 2 # To generate two diverse summaries
num_beam_groups = 2
model_kwargs = {
"encoder_outputs": model.get_encoder()(
@ -1189,7 +1191,7 @@ class HammingDiversityLogitsProcessor(LogitsProcessor):
MinLengthLogitsProcessor(10, eos_token_id=model.config.eos_token_id),
]
)
#generate the diverse summary using group_beam_search
outputs_diverse = model.group_beam_search(
encoder_input_ids.repeat_interleave(num_beams, dim=0), beam_scorer, logits_processor=logits_processor_diverse, **model_kwargs
)