Make tests run in less time by reducing batch_size

.circleci/create_circleci_config.py

@@ -187,6 +187,11 @@ def to_dict(self):
             # If it's done already, the files are inside the directory `/test_data/`.
             {"run": {"name": "fetch hub objects before pytest", "command": "cp -r /test_data/* . 2>/dev/null || true; python3 utils/fetch_hub_objects_for_ci.py"}},
             {"run": {"name": "download and unzip hub cache", "command": 'curl -L -o huggingface-cache.tar.gz https://huggingface.co/datasets/hf-internal-testing/hf_hub_cache/resolve/main/huggingface-cache.tar.gz && apt-get install pigz && tar --use-compress-program="pigz -d -p 8" -xf huggingface-cache.tar.gz && mv -n hub/* /root/.cache/huggingface/hub/ && ls -la /root/.cache/huggingface/hub/'}},
+            {"run": {
+                "name": "git checkout main",
+                "command": "git fetch origin && git checkout main && git pull origin main",
+            }
+            },
             {"run": {
                 "name": "Run tests",
                 "command": f"({timeout_cmd} python3 -m pytest {marker_cmd} -n {self.pytest_num_workers} {junit_flags} {repeat_on_failure_flags} {' '.join(pytest_flags)} $(cat splitted_tests.txt) | tee tests_output.txt)"}

tests/models/albert/test_modeling_albert.py

@@ -46,7 +46,7 @@ class AlbertModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_input_mask=True,

tests/models/aria/test_modeling_aria.py

@@ -60,7 +60,7 @@ class AriaVisionText2TextModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         num_channels=3,
         image_size=16,
         num_image_tokens=4,

tests/models/audio_spectrogram_transformer/test_modeling_audio_spectrogram_transformer.py

@@ -45,7 +45,7 @@ class ASTModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         patch_size=2,
         max_length=24,
         num_mel_bins=16,

tests/models/autoformer/test_modeling_autoformer.py

@@ -43,7 +43,7 @@ def __init__(
         self,
         parent,
         d_model=16,
-        batch_size=13,
+        batch_size=2,
         prediction_length=7,
         context_length=14,
         label_length=10,

tests/models/bamba/test_modeling_bamba.py

@@ -64,7 +64,7 @@ class BambaModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_input_mask=True,

tests/models/bart/test_modeling_bart.py

@@ -76,7 +76,7 @@ class BartModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_labels=False,
@@ -1313,7 +1313,7 @@ def __init__(
         self,
         parent,
         vocab_size=99,
-        batch_size=13,
+        batch_size=2,
         d_model=16,
         decoder_seq_length=7,
         is_training=True,

tests/models/beit/test_modeling_beit.py

@@ -63,8 +63,8 @@ def __init__(
         self,
         parent,
         vocab_size=100,
-        batch_size=13,
-        image_size=30,
+        batch_size=2,
+        image_size=8,
         patch_size=2,
         num_channels=3,
         is_training=True,

tests/models/bert/test_modeling_bert.py

@@ -54,7 +54,7 @@ class BertModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_input_mask=True,

tests/models/bert_generation/test_modeling_bert_generation.py

@@ -36,7 +36,7 @@ class BertGenerationEncoderTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_input_mask=True,

tests/models/big_bird/test_modeling_big_bird.py

@@ -45,8 +45,8 @@ class BigBirdModelTester:
     def __init__(
         self,
         parent,
-        batch_size=7,
-        seq_length=128,
+        batch_size=2,
+        seq_length=32,
         is_training=True,
         use_input_mask=True,
         use_token_type_ids=True,

tests/models/bigbird_pegasus/test_modeling_bigbird_pegasus.py

@@ -78,8 +78,8 @@ class BigBirdPegasusModelTester:
     def __init__(
         self,
         parent,
-        batch_size=7,
-        seq_length=256,
+        batch_size=2,
+        seq_length=64,
         is_training=True,
         use_labels=False,
         vocab_size=99,

tests/models/biogpt/test_modeling_biogpt.py

@@ -41,7 +41,7 @@ class BioGptModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_input_mask=True,

tests/models/bitnet/test_modeling_bitnet.py

@@ -43,7 +43,7 @@ class BitNetModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_input_mask=True,

tests/models/blenderbot_small/test_modeling_blenderbot_small.py

@@ -66,7 +66,7 @@ class BlenderbotSmallModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_labels=False,
@@ -339,7 +339,7 @@ def __init__(
         self,
         parent,
         vocab_size=99,
-        batch_size=13,
+        batch_size=2,
         d_model=16,
         decoder_seq_length=7,
         is_training=True,

tests/models/bros/test_modeling_bros.py

@@ -40,7 +40,7 @@ class BrosModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_input_mask=True,

tests/models/canine/test_modeling_canine.py

@@ -39,7 +39,7 @@ class CanineModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_input_mask=True,

tests/models/chameleon/test_modeling_chameleon.py

@@ -51,7 +51,7 @@ class ChameleonModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=35,
         is_training=False,
         use_input_mask=True,

tests/models/chinese_clip/test_modeling_chinese_clip.py

@@ -55,7 +55,7 @@ class ChineseCLIPTextModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_input_mask=True,

tests/models/cohere/test_modeling_cohere.py

@@ -46,7 +46,7 @@ class CohereModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_input_mask=True,

tests/models/convbert/test_modeling_convbert.py

@@ -42,7 +42,7 @@ class ConvBertModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_input_mask=True,

tests/models/convnextv2/test_modeling_convnextv2.py

@@ -43,7 +43,7 @@ class ConvNextV2ModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         image_size=32,
         num_channels=3,
         num_stages=4,

tests/models/cvt/test_modeling_cvt.py

@@ -49,7 +49,7 @@ class CvtModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         image_size=64,
         num_channels=3,
         embed_dim=[16, 32, 48],

tests/models/data2vec/test_modeling_data2vec_audio.py

@@ -46,7 +46,7 @@ class Data2VecAudioModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=1024,  # speech is longer
         is_training=False,
         hidden_size=16,

tests/models/data2vec/test_modeling_data2vec_text.py

@@ -47,7 +47,7 @@ class Data2VecTextModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_input_mask=True,

tests/models/data2vec/test_modeling_data2vec_vision.py

@@ -59,7 +59,7 @@ def __init__(
         self,
         parent,
         vocab_size=100,
-        batch_size=13,
+        batch_size=2,
         image_size=30,
         patch_size=2,
         num_channels=3,

tests/models/deberta/test_modeling_deberta.py

@@ -37,7 +37,7 @@ class DebertaModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_input_mask=True,

tests/models/deberta_v2/test_modeling_deberta_v2.py

@@ -38,7 +38,7 @@ class DebertaV2ModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_input_mask=True,

tests/models/decision_transformer/test_modeling_decision_transformer.py

@@ -34,7 +34,7 @@ class DecisionTransformerModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         act_dim=6,
         state_dim=17,

tests/models/deepseek_v3/test_modeling_deepseek_v3.py

@@ -52,7 +52,7 @@ class DeepseekV3ModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_input_mask=True,

tests/models/deformable_detr/test_modeling_deformable_detr.py

@@ -50,7 +50,7 @@ class DeformableDetrModelTester:
     def __init__(
         self,
         parent,
-        batch_size=8,
+        batch_size=2,
         is_training=True,
         use_labels=True,
         hidden_size=32,

tests/models/deit/test_modeling_deit.py

@@ -61,8 +61,8 @@ class DeiTModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
-        image_size=30,
+        batch_size=2,
+        image_size=8,
         patch_size=2,
         num_channels=3,
         is_training=True,

tests/models/diffllama/test_modeling_diffllama.py

@@ -56,7 +56,7 @@ class DiffLlamaModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_input_mask=True,

tests/models/dinat/test_modeling_dinat.py

@@ -43,7 +43,7 @@ class DinatModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         image_size=64,
         patch_size=4,
         num_channels=3,

tests/models/dinov2/test_modeling_dinov2.py

@@ -48,7 +48,7 @@ class Dinov2ModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         image_size=30,
         patch_size=2,
         num_channels=3,

tests/models/dinov2_with_registers/test_modeling_dinov2_with_registers.py

@@ -52,7 +52,7 @@ class Dinov2WithRegistersModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         image_size=30,
         patch_size=2,
         num_channels=3,

tests/models/dinov3_convnext/test_modeling_dinov3_convnext.py

@@ -42,7 +42,7 @@ class DINOv3ConvNextModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         image_size=32,
         num_channels=3,
         hidden_sizes=[10, 20, 30, 40],

tests/models/dinov3_vit/test_modeling_dinov3_vit.py

@@ -42,7 +42,7 @@ class DINOv3ViTModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         image_size=30,
         patch_size=2,
         num_channels=3,

tests/models/distilbert/test_modeling_distilbert.py

@@ -44,7 +44,7 @@ class DistilBertModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=True,
         use_input_mask=True,

tests/models/donut/test_modeling_donut_swin.py

@@ -36,7 +36,7 @@ class DonutSwinModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         image_size=32,
         patch_size=2,
         num_channels=3,

tests/models/dpr/test_modeling_dpr.py

@@ -34,7 +34,7 @@ class DPRModelTester:
     def __init__(
         self,
         parent,
-        batch_size=13,
+        batch_size=2,
         seq_length=7,
         is_training=False,
         use_input_mask=True,