Merge pull request #1830 from AI-Hypercomputer:model_comment

PiperOrigin-RevId: 771232937

Author: maxtext authors

benchmarks/maxtext_trillium_model_configs.py

@@ -722,7 +722,7 @@
     ),
 )
 
-
+# Config only runs on v6e-256
 llama3_1_405b_8192_fsdp_dcn = _add_to_model_dictionary(
     trillium_model_dict,
     MaxTextModel(
@@ -776,6 +776,7 @@
     ),
 )
 
+# Config only runs on v6e-256
 llama3_1_405b_8192_pure_fsdp_ici = _add_to_model_dictionary(
   trillium_model_dict,
   MaxTextModel(
@@ -1524,6 +1525,7 @@
     ),
 )
 
+# Config only runs on v6e-256
 mixtral_8x22b_dropped = _add_to_model_dictionary(
     trillium_model_dict,
     MaxTextModel(
@@ -1568,6 +1570,7 @@
     ),
 )
 
+# Config only runs on v6e-256
 deepseek_v3_ep16 = _add_to_model_dictionary(
     trillium_model_dict,
     MaxTextModel(
@@ -1608,7 +1611,7 @@
     ),
 )
 
-
+# Config only runs on v6e-256
 gemma2_9b_8192 = _add_to_model_dictionary(
     trillium_model_dict,
     MaxTextModel(
@@ -1641,7 +1644,7 @@
     ),
 )
 
-
+# Config only runs on v6e-256
 gemma2_27b_8192 = _add_to_model_dictionary(
   trillium_model_dict,
   MaxTextModel(
@@ -1736,7 +1739,7 @@
   )
 )
 
-# Customized MoE model - 700B
+# Customized MoE model - 700B, and config only runs on v6e-256
 custom_moe_700b = _add_to_model_dictionary(
     trillium_model_dict,
     MaxTextModel(
@@ -1797,5 +1800,3 @@
         ),
     ),
 )
-
-