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

* CI with layers=2

---------
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

tests/models/electra/test_modeling_electra.py

@@ -54,7 +54,7 @@ class ElectraModelTester:
         use_labels=True,
         vocab_size=99,
         hidden_size=32,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_size=37,
         hidden_act="gelu",

tests/models/electra/test_modeling_flax_electra.py

@@ -34,7 +34,7 @@ class FlaxElectraModelTester(unittest.TestCase):
         vocab_size=99,
         embedding_size=24,
         hidden_size=32,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_size=37,
         hidden_act="gelu",

tests/models/ernie/test_modeling_ernie.py

@@ -56,7 +56,7 @@ class ErnieModelTester:
         use_labels=True,
         vocab_size=99,
         hidden_size=32,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_size=37,
         hidden_act="gelu",

tests/models/ernie_m/test_modeling_ernie_m.py

@@ -50,7 +50,7 @@ class ErnieMModelTester:
         use_labels=True,
         vocab_size=99,
         hidden_size=32,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_size=37,
         hidden_act="gelu",

tests/models/esm/test_modeling_esm.py

@@ -49,7 +49,7 @@ class EsmModelTester:
         use_labels=True,
         vocab_size=33,
         hidden_size=32,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_size=37,
         hidden_act="gelu",

tests/models/esm/test_modeling_esmfold.py

@@ -43,7 +43,7 @@ class EsmFoldModelTester:
         use_labels=False,
         vocab_size=19,
         hidden_size=32,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_size=37,
         hidden_act="gelu",

tests/models/falcon/test_modeling_falcon.py

@@ -50,7 +50,7 @@ class FalconModelTester:
         use_labels=True,
         vocab_size=99,
         hidden_size=32,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_size=37,
         hidden_act="gelu",

tests/models/flaubert/test_modeling_flaubert.py

@@ -57,7 +57,7 @@ class FlaubertModelTester(object):
         vocab_size=99,
         n_special=0,
         hidden_size=32,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         num_attention_heads=4,
         hidden_dropout_prob=0.1,
         attention_probs_dropout_prob=0.1,

tests/models/flava/test_modeling_flava.py

@@ -79,7 +79,7 @@ class FlavaImageModelTester:
         parent,
         batch_size=12,
         hidden_size=32,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_size=37,
         hidden_act="gelu",
@@ -342,7 +342,7 @@ class FlavaTextModelTester:
         max_position_embeddings=512,
         position_embedding_type="absolute",
         hidden_size=32,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_size=37,
         hidden_act="gelu",
@@ -487,7 +487,7 @@ class FlavaMultimodalModelTester:
         seq_length=44,
         use_input_mask=True,
         hidden_size=32,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_size=37,
         hidden_act="gelu",

tests/models/fnet/test_modeling_fnet.py

@@ -70,7 +70,7 @@ class FNetModelTester:
         use_labels=True,
         vocab_size=99,
         hidden_size=32,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         intermediate_size=37,
         hidden_act="gelu",
         hidden_dropout_prob=0.1,

tests/models/git/test_modeling_git.py

@@ -51,7 +51,7 @@ class GitVisionModelTester:
         is_training=True,
         hidden_size=32,
         projection_dim=32,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_size=37,
         dropout=0.1,
@@ -203,7 +203,7 @@ class GitModelTester:
         use_labels=True,
         vocab_size=99,
         hidden_size=32,
-        num_hidden_layers=4,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_size=37,
         hidden_act="gelu",

tests/models/gpt2/test_modeling_flax_gpt2.py

@@ -52,7 +52,7 @@ class FlaxGPT2ModelTester:
         use_labels=True,
         vocab_size=99,
         hidden_size=32,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_size=37,
         hidden_act="gelu",

tests/models/gpt2/test_modeling_gpt2.py

@@ -56,7 +56,7 @@ class GPT2ModelTester:
         use_mc_token_ids=True,
         vocab_size=99,
         hidden_size=32,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_size=37,
         hidden_act="gelu",

tests/models/gpt_bigcode/test_modeling_gpt_bigcode.py

@@ -55,7 +55,7 @@ class GPTBigCodeModelTester:
         use_mc_token_ids=True,
         vocab_size=99,
         hidden_size=32,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_size=37,
         hidden_act="relu",

tests/models/gpt_neo/test_modeling_flax_gpt_neo.py

@@ -52,9 +52,9 @@ class FlaxGPTNeoModelTester:
         use_labels=True,
         vocab_size=99,
         hidden_size=32,
-        num_hidden_layers=4,
+        num_hidden_layers=2,
         num_attention_heads=4,
-        attention_types=[[["global", "local"], 2]],
+        attention_types=[[["global", "local"], 1]],
         intermediate_size=37,
         hidden_act="gelu",
         hidden_dropout_prob=0.1,

tests/models/gpt_neo/test_modeling_gpt_neo.py

@@ -54,8 +54,8 @@ class GPTNeoModelTester:
         use_mc_token_ids=True,
         vocab_size=99,
         hidden_size=32,
-        num_hidden_layers=4,
-        attention_types=[[["global", "local"], 2]],
+        num_hidden_layers=2,
+        attention_types=[[["global", "local"], 1]],
         num_attention_heads=4,
         intermediate_size=37,
         hidden_act="gelu",

tests/models/gpt_neox/test_modeling_gpt_neox.py

@@ -52,7 +52,7 @@ class GPTNeoXModelTester:
         use_labels=True,
         vocab_size=99,
         hidden_size=64,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_size=37,
         hidden_act="gelu",

tests/models/gpt_neox_japanese/test_modeling_gpt_neox_japanese.py

@@ -44,7 +44,7 @@ class GPTNeoXJapaneseModelTester:
         use_labels=True,
         vocab_size=99,
         hidden_size=32,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_multiple_size=4,
         hidden_act="gelu",

tests/models/gptj/test_modeling_flax_gptj.py

@@ -53,7 +53,7 @@ class FlaxGPTJModelTester:
         vocab_size=99,
         hidden_size=32,
         rotary_dim=4,
-        num_hidden_layers=4,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_size=37,
         hidden_act="gelu",

tests/models/gptj/test_modeling_gptj.py

@@ -56,7 +56,7 @@ class GPTJModelTester:
         vocab_size=99,
         hidden_size=32,
         rotary_dim=4,
-        num_hidden_layers=5,
+        num_hidden_layers=2,
         num_attention_heads=4,
         intermediate_size=37,
         hidden_act="gelu",