Updated NPI files gemma3-27B (#567)

Updated NPI files to get good output in AIC.

Signed-off-by: Ann Kuruvilla <[email protected]>

Author: quic-akuruvil

examples/gemma3_example/fp32_nodes_gemma3_27b_image.yaml renamed to examples/gemma3_example/fp32_nodes_gemma3_27b.yaml

@@ -11,6 +11,7 @@
 
 from QEfficient import QEFFAutoModelForImageTextToText
 
+# Change model_id to "google/gemma-3-27b-it" for 27B model
 model_id = "google/gemma-3-4b-it"
 config = AutoConfig.from_pretrained(model_id)
 # For Testing Purpose Only
@@ -25,8 +26,8 @@
     model_id, config=config, attn_implementation="eager", kv_offload=True
 )
 
-### use skip_vision=Ture, if want to run only text, or false ###
-skip_vision = True
+### use skip_vision=True, if want to run only text, or false ###
+skip_vision = False
 
 if skip_vision:
     ## Only Text ##
@@ -41,7 +42,7 @@
         aic_enable_depth_first=True,
         skip_vision=True,
         mos=1,
-        node_precision_info="examples/gemma3_example/fp32_mm.yaml",
+        node_precision_info="examples/gemma3_example/fp32_nodes_gemma3_4b.yaml",  # Change to fp32_nodes_gemma3_27b.yaml for 27B model
     )
 
     messages = [
@@ -77,7 +78,7 @@
         mxint8_kv_cache=False,
         aic_enable_depth_first=True,
         mos=1,
-        node_precision_info="examples/gemma3_example/fp32_mm.yaml",
+        node_precision_info="examples/gemma3_example/fp32_nodes_gemma3_4b.yaml",  # Change to fp32_nodes_gemma3_27b.yaml for 27B model
     )
 
     ### IMAGE + TEXT ###