From 0896a3bc11fdb5da5d58a97dcb753ae70781820a Mon Sep 17 00:00:00 2001
From: loyce-cheng <129246079+loyce-cheng@users.noreply.github.com>
Date: Sat, 3 Oct 2026 02:36:13 +0800
Subject: [PATCH] fix(huggingface): preserve batching when pad_token_id is zero

---
 .../mlserver_huggingface/common.py            |  2 +-
 runtimes/huggingface/tests/test_common.py     | 46 ++++++++++++++++++-
 2 files changed, 45 insertions(+), 3 deletions(-)

diff --git a/runtimes/huggingface/mlserver_huggingface/common.py b/runtimes/huggingface/mlserver_huggingface/common.py
index e53d7f6e2..39d58ad62 100644
--- a/runtimes/huggingface/mlserver_huggingface/common.py
+++ b/runtimes/huggingface/mlserver_huggingface/common.py
@@ -63,7 +63,7 @@ def load_pipeline_from_settings(
     # If max_batch_size > 1 we need to ensure tokens are padded
     if settings.max_batch_size > 1:
         model = hf_pipeline.model
-        if not hf_pipeline.tokenizer.pad_token_id:
+        if hf_pipeline.tokenizer.pad_token_id is None:
             eos_token_id = model.config.eos_token_id  # type: ignore
             if eos_token_id:
                 hf_pipeline.tokenizer.pad_token_id = [str(eos_token_id)]  # type: ignore
diff --git a/runtimes/huggingface/tests/test_common.py b/runtimes/huggingface/tests/test_common.py
index 894570842..a473a3318 100644
--- a/runtimes/huggingface/tests/test_common.py
+++ b/runtimes/huggingface/tests/test_common.py
@@ -243,8 +243,8 @@ def test_pipeline_cpu_device_set(
             "hf-internal-testing/tiny-bert-for-token-classification",
             "token-classification",
             10,
-            1,
-        ),  # Neither pad_token nor eos_token defined revert to 1
+            10,
+        ),
     ],
 )
 def test_pipeline_checks_for_eos_and_pad_token(
@@ -265,3 +265,45 @@ def test_pipeline_checks_for_eos_and_pad_token(
     m = load_pipeline_from_settings(hf_settings, model_settings)
 
     assert m._batch_size == expected_batch_size
+
+
+@pytest.mark.parametrize("pad_token_id", [0, 1])
+@pytest.mark.parametrize("eos_token_id", [None, 2])
+@patch("mlserver_huggingface.common._get_pipeline_class")
+def test_pipeline_preserves_configured_padding(
+    mock_pipeline_factory, pad_token_id: int, eos_token_id: Optional[int], caplog
+):
+    hf_settings = HuggingFaceSettings(pretrained_model="some-model")
+    model_settings = ModelSettings(
+        name="foo", implementation=HuggingFaceRuntime, max_batch_size=8
+    )
+    mock_pipeline = mock_pipeline_factory.return_value.return_value
+    mock_pipeline.tokenizer.pad_token_id = pad_token_id
+    mock_pipeline.model.config.eos_token_id = eos_token_id
+    mock_pipeline._batch_size = model_settings.max_batch_size
+
+    pipeline = load_pipeline_from_settings(hf_settings, model_settings)
+
+    assert pipeline._batch_size == model_settings.max_batch_size
+    assert pipeline.tokenizer.pad_token_id == pad_token_id
+    assert "setting batch size to 1" not in caplog.text
+
+
+@patch("mlserver_huggingface.common._get_pipeline_class")
+def test_pipeline_without_padding_or_eos_uses_single_batch(
+    mock_pipeline_factory, caplog
+):
+    hf_settings = HuggingFaceSettings(pretrained_model="some-model")
+    model_settings = ModelSettings(
+        name="foo", implementation=HuggingFaceRuntime, max_batch_size=8
+    )
+    mock_pipeline = mock_pipeline_factory.return_value.return_value
+    mock_pipeline.tokenizer.pad_token_id = None
+    mock_pipeline.model.config.eos_token_id = None
+    mock_pipeline._batch_size = model_settings.max_batch_size
+
+    pipeline = load_pipeline_from_settings(hf_settings, model_settings)
+
+    assert pipeline._batch_size == 1
+    assert pipeline.tokenizer.pad_token_id is None
+    assert "setting batch size to 1" in caplog.text