SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

CI with num_hidden_layers=2 🚀🚀🚀 (#25266)

Browse Source 
* CI with layers=2

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
This commit is contained in:
Yih-Dar
2023-08-02 20:22:36 +02:00
committed by GitHub
parent b28ebb2655
commit bd90cda9a6
147 changed files with 207 additions and 196 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
tests/models/prophetnet/test_modeling_prophetnet.py
View File

@@ -55,10 +55,10 @@ class ProphetNetModelTester:
use_labels=True,
decoder_start_token_id=0,
encoder_ffn_dim=32,
num_encoder_layers=4,
num_encoder_layers=2,
num_encoder_attention_heads=4,
decoder_ffn_dim=32,
num_decoder_layers=4,
num_decoder_layers=2,
num_decoder_attention_heads=4,
max_position_embeddings=30,
is_encoder_decoder=True,
@@ -437,10 +437,10 @@ class ProphetNetModelTester:
decoder_attention_mask=decoder_attention_mask,
labels=lm_labels,
)
self.parent.assertTrue(torch.allclose(result.loss, torch.tensor(4.5981, device=torch_device), atol=1e-3))
self.parent.assertTrue(torch.allclose(result.loss, torch.tensor(4.5892, device=torch_device), atol=1e-3))
expected_logit_slice = torch.tensor(
[-0.0648, 0.0790, 0.0360, 0.0089, 0.0039, -0.0639, 0.0131], device=torch_device
[-0.0184, 0.0758, -0.0543, -0.0093, 0.0050, -0.0660, -0.1453], device=torch_device
)
self.parent.assertTrue(torch.allclose(result.logits[0, :, 1], expected_logit_slice, atol=1e-3))
@@ -551,10 +551,10 @@ class ProphetNetStandaloneDecoderModelTester:
use_labels=True,
decoder_start_token_id=0,
encoder_ffn_dim=32,
num_encoder_layers=4,
num_encoder_layers=2,
num_encoder_attention_heads=4,
decoder_ffn_dim=32,
num_decoder_layers=4,
num_decoder_layers=2,
num_decoder_attention_heads=4,
max_position_embeddings=30,
is_encoder_decoder=False,
@@ -782,10 +782,10 @@ class ProphetNetStandaloneEncoderModelTester:
use_labels=True,
decoder_start_token_id=0,
encoder_ffn_dim=32,
num_encoder_layers=4,
num_encoder_layers=2,
num_encoder_attention_heads=4,
decoder_ffn_dim=32,
num_decoder_layers=4,
num_decoder_layers=2,
num_decoder_attention_heads=4,
max_position_embeddings=30,
is_encoder_decoder=False,