diff --git a/docs/docs/advanced_features/server_arguments.mdx b/docs/docs/advanced_features/server_arguments.mdx
index 897e1dc83..fccc06f5a 100644
--- a/docs/docs/advanced_features/server_arguments.mdx
+++ b/docs/docs/advanced_features/server_arguments.mdx
@@ -519,6 +519,12 @@ Please consult the documentation below and [server_args.py](https://github.com/s
| How conservative the schedule policy is. A larger value means more conservative scheduling. Use a larger value if you see requests being retracted frequently. |
`1.0` |
Type: float |
+
+
+ | `--retraction-policy` |
+ The decode retraction policy to use when the KV cache is full. 'length' preserves the existing behavior and retracts short-output, long-input requests first. 'priority' retracts lower-priority requests first, using the same priority direction as priority scheduling. |
+ `length` |
+ length, priority |
| `--page-size` |
@@ -2763,6 +2769,12 @@ Please consult the documentation below and [server_args.py](https://github.com/s
`False` |
bool flag (set to enable) |
+
+ | `--return-hidden-states-mode` |
+ Set the maximum hidden-state return mode supported by the server. `last` allows requests with return_hidden_states=False or `last`; `full` also allows return_hidden_states=True. |
+ `None` |
+ last, full |
+
| `--enable-return-routed-experts` |
Enable returning routed experts of each layer with responses. |
@@ -3106,6 +3118,12 @@ Please consult the documentation below and [server_args.py](https://github.com/s
For VLM, load weights for the language model only. |
`False` |
bool flag (set to enable) |
+
+
+ | `--language-model-only` |
+ Skip the multimodal encoder entirely: its weights are never loaded and the tower is never built, freeing that GPU memory for KV cache. Multimodal requests are rejected. Unlike --language-only this is a standalone mode, not part of encoder/decoder disaggregation. |
+ `False` |
+ bool flag (set to enable) |
| `--encoder-transfer-backend` |