From b16d446827b110b15aa09f3edb4fb51ce90c9a95 Mon Sep 17 00:00:00 2001
From: Colman Glagovich <cglagovich@tenstorrent.com>
Date: Fri, 6 Dec 2024 11:54:12 -0800
Subject: [PATCH] #0: Fix CI: Don't cache weights for quick, AllPostCommit
 Llama test

---
 models/demos/llama3/tt/llama_embedding.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/models/demos/llama3/tt/llama_embedding.py b/models/demos/llama3/tt/llama_embedding.py
index a89ac40db9a..672f0b980f1 100644
--- a/models/demos/llama3/tt/llama_embedding.py
+++ b/models/demos/llama3/tt/llama_embedding.py
@@ -23,7 +23,7 @@ def __init__(
 
         base_name = args.get_state_dict_prefix("", None) + "tok_embeddings.weight"
         torch_weight = self.state_dict[base_name].unsqueeze(0).unsqueeze(0)
-        cache_name = weight_cache_path / base_name
+        cache_name = None if args.dummy_weights else weight_cache_path / base_name
         self.weights = ttnn.as_tensor(
             torch_weight,
             dtype=dtype,