From 91e468bfadbb82beaee0191b8c9067c4e3859933 Mon Sep 17 00:00:00 2001 From: Salar Hosseini Date: Tue, 17 Dec 2024 16:20:06 +0000 Subject: [PATCH] fix typo Signed-off-by: Salar Hosseini --- models/demos/llama3/tt/generator_vllm.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/models/demos/llama3/tt/generator_vllm.py b/models/demos/llama3/tt/generator_vllm.py index bde347e30677..7989aba9547b 100644 --- a/models/demos/llama3/tt/generator_vllm.py +++ b/models/demos/llama3/tt/generator_vllm.py @@ -38,7 +38,7 @@ def input_processor_for_mllama(ctx: InputContext, inputs: Union[DecoderOnlyInput inputs["encoder_multi_modal_data"] = {} return inputs - # Set encoder prompt length based on the number of vision tokens so block manager allocates enable blocks (cross block tables). + # Set encoder prompt length based on the number of vision tokens so block manager allocates enough blocks (cross block tables). hf_config = ctx.model_config.hf_config assert hf_config.vision_config.image_size % 14 == 0, "chunk size should be multiple of 14" token_per_chunk = nearest_32(