diff --git a/alfred/fm/huggingface.py b/alfred/fm/huggingface.py
index 54d42bd..a989243 100644
--- a/alfred/fm/huggingface.py
+++ b/alfred/fm/huggingface.py
@@ -357,6 +357,7 @@ def _generate_batch(
             inputs = batch
         else:
             if padding:
+                self.tokenizer.pad_token = self.tokenizer.eos_token
                 inputs = self.tokenizer(batch,
                                         return_tensors="pt",
                                         padding=True)
diff --git a/docs/alfred/fm/vllm.md b/docs/alfred/fm/vllm.md
new file mode 100644
index 0000000..497417b
--- /dev/null
+++ b/docs/alfred/fm/vllm.md
@@ -0,0 +1,35 @@
+# Vllm
+
+[Alfred Index](../../README.md#alfred-index) /
+[Alfred](../index.md#alfred) /
+[Fm](./index.md#fm) /
+Vllm
+
+> Auto-generated documentation for [alfred.fm.vllm](../../../alfred/fm/vllm.py) module.
+
+- [Vllm](#vllm)
+  - [vLLMModel](#vllmmodel)
+
+## vLLMModel
+
+[Show source in vllm.py:18](../../../alfred/fm/vllm.py#L18)
+
+vLLMModel wraps a vLLM model. vLLM is a fast and easy-to-use library for LLM inference.
+
+source: https://github.com/vllm-project/vllm
+
+#### Signature
+
+```python
+class vLLMModel(LocalAccessFoundationModel):
+    def __init__(
+        self, model: str, model_string: str, local_dir: str = None, **kwargs: Any
+    ):
+        ...
+```
+
+#### See also
+
+- [LocalAccessFoundationModel](./model.md#localaccessfoundationmodel)
+
+