[
https://issues.apache.org/jira/browse/BEAM-14337?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Anand Inguva updated BEAM-14337:
--------------------------------
Description:
Some models in Pytorch instantiating from torch.nn.Module, has extra parameters
in the forward function call. These extra parameters can be passed as Dict or
as positional arguments.
Example of PyTorch models supported by Hugging Face ->
[https://huggingface.co/bert-base-uncased]
[Some torch models on Hugging
face|https://github.com/huggingface/transformers/blob/main/src/transformers/models/bert/modeling_bert.py]
Eg:
[https://huggingface.co/docs/transformers/model_doc/bert#transformers.BertModel]
{code:java}
inputs = {
input_ids: Tensor1,
attention_mask: Tensor2,
token_type_ids: Tensor3,
}
model = BertModel.from_pretrained("bert-base-uncased") # which is a #subclass
of torch.nn.Module
outputs = model(**inputs) # model forward method should be expecting the keys
in the inputs as the positional arguments.{code}
was:
Some models in Pytorch instantiating from torch.nn.Module, has extra parameters
in the forward function call. These extra parameters can be passed as Dict or
as positional arguments.
Example of PyTorch models supported by Hugging Face ->
https://huggingface.co/bert-base-uncased
[Some torch models on Hugging
face|https://github.com/huggingface/transformers/blob/main/src/transformers/models/bert/modeling_bert.py]
Eg:
https://huggingface.co/docs/transformers/model_doc/bert#transformers.BertModel
inputs = {
input_ids: Tensor1,
attention_mask: Tensor2,
token_type_ids: Tensor3,
}
model = BertModel.from_pretrained("bert-base-uncased") # which is a subclass of
#
{code:java}
inputs = {
input_ids: Tensor1,
attention_mask: Tensor2,
token_type_ids: Tensor3,
}
model = BertModel.from_pretrained("bert-base-uncased") # which is a #subclass
of torch.nn.Module{code}
torch.nn.Module.
> Support **kwargs for PyTorch models.
> ------------------------------------
>
> Key: BEAM-14337
> URL: https://issues.apache.org/jira/browse/BEAM-14337
> Project: Beam
> Issue Type: Sub-task
> Components: sdk-py-core
> Reporter: Anand Inguva
> Priority: P2
>
> Some models in Pytorch instantiating from torch.nn.Module, has extra
> parameters in the forward function call. These extra parameters can be passed
> as Dict or as positional arguments.
> Example of PyTorch models supported by Hugging Face ->
> [https://huggingface.co/bert-base-uncased]
> [Some torch models on Hugging
> face|https://github.com/huggingface/transformers/blob/main/src/transformers/models/bert/modeling_bert.py]
> Eg:
> [https://huggingface.co/docs/transformers/model_doc/bert#transformers.BertModel]
> {code:java}
> inputs = {
> input_ids: Tensor1,
> attention_mask: Tensor2,
> token_type_ids: Tensor3,
> }
> model = BertModel.from_pretrained("bert-base-uncased") # which is a
> #subclass of torch.nn.Module
> outputs = model(**inputs) # model forward method should be expecting the keys
> in the inputs as the positional arguments.{code}
>
>
>
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
