A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 4, 06:15 UTC
[Bugfix] Suggest upgrading Transformers for tokenizer class errors (#38099)
Signed-off-by: Lidang-Jiang <lidangjiang@gmail.com>
Lidang Jiang committed
May 5, 2026 at 22:10 UTC
b786ec8e744b0c1a48b58873a14adb4e4f30eec3
1 file changed
+4
-1
vllm/tokenizers/hf.py
+4
-1
@@ -190,7 +190,10 @@ class CachedHfTokenizer(TokenizerLike):
190
"is a custom tokenizer not yet available in the "
191
"HuggingFace transformers library, consider "
192
"setting `trust_remote_code=True` in LLM or using "
193
- "the `--trust-remote-code` flag in the CLI."
193
+ "the `--trust-remote-code` flag in the CLI. If the "
194
+ "model was created with a newer version of "
195
+ "transformers, consider upgrading: "
196
+ "`uv pip install --upgrade transformers`"
197
)
198
raise RuntimeError(err_msg) from e
199
else: