Fix scripts

Signed-off-by: Onur Yilmaz <oyilmaz@nvidia.com>

Author: oyilmaz-nvidia

scripts/deploy/llm/automodel/deploy_ray.py

@@ -96,7 +96,7 @@ def parse_args():
     parser.add_argument(
         "--model_id",
         type=str,
-        default="nemo-model",
+        default="megatron-model",
         help="Identifier for the model in the API responses",
     )
     parser.add_argument(
@@ -167,12 +167,6 @@ def parse_args():
         default=None,
         help="Random seed for reproducible inference",
     )
-    parser.add_argument(
-        "--megatron_checkpoint",
-        type=str,
-        default=None,
-        help="Path to the Megatron checkpoint file",
-    )
     parser.add_argument(
         "--model_type",
         type=str,
@@ -206,12 +200,6 @@ def main():
         port=args.port,
         runtime_env=runtime_env,
     )
-    if args.nemo_checkpoint:
-        model_format = "nemo"
-    elif args.megatron_checkpoint:
-        model_format = "megatron"
-    else:
-        raise ValueError("Either --nemo_checkpoint or --megatron_checkpoint must be provided")
 
     model_config_kwargs = {
         "account_for_embedding_in_pipeline_split": args.account_for_embedding_in_pipeline_split,
@@ -240,9 +228,7 @@ def main():
         legacy_ckpt=args.legacy_ckpt,
         max_batch_size=args.max_batch_size,
         random_seed=args.random_seed,
-        megatron_checkpoint_filepath=args.megatron_checkpoint,
         model_type=args.model_type,
-        model_format=model_format,
         micro_batch_size=args.micro_batch_size,
         **model_config_kwargs,
     )

scripts/deploy/llm/mbridge/deploy_ray.py

@@ -96,7 +96,7 @@ def parse_args():
     parser.add_argument(
         "--model_id",
         type=str,
-        default="nemo-model",
+        default="megatron-model",
         help="Identifier for the model in the API responses",
     )
     parser.add_argument(
@@ -167,12 +167,6 @@ def parse_args():
         default=None,
         help="Random seed for reproducible inference",
     )
-    parser.add_argument(
-        "--megatron_checkpoint",
-        type=str,
-        default=None,
-        help="Path to the Megatron checkpoint file",
-    )
     parser.add_argument(
         "--model_type",
         type=str,
@@ -206,12 +200,6 @@ def main():
         port=args.port,
         runtime_env=runtime_env,
     )
-    if args.nemo_checkpoint:
-        model_format = "nemo"
-    elif args.megatron_checkpoint:
-        model_format = "megatron"
-    else:
-        raise ValueError("Either --nemo_checkpoint or --megatron_checkpoint must be provided")
 
     model_config_kwargs = {
         "account_for_embedding_in_pipeline_split": args.account_for_embedding_in_pipeline_split,
@@ -240,9 +228,7 @@ def main():
         legacy_ckpt=args.legacy_ckpt,
         max_batch_size=args.max_batch_size,
         random_seed=args.random_seed,
-        megatron_checkpoint_filepath=args.megatron_checkpoint,
         model_type=args.model_type,
-        model_format=model_format,
         micro_batch_size=args.micro_batch_size,
         **model_config_kwargs,
     )

scripts/deploy/llm/mlm/deploy_ray.py

@@ -209,13 +209,6 @@ def get_args(argv):
         default="gpt",
         help="Type of model to load",
     )
-    parser.add_argument(
-        "-mf",
-        "--model_format",
-        type=str,
-        default="nemo",
-        help="Format of model to load",
-    )
     parser.add_argument(
         "-mb",
         "--micro_batch_size",

scripts/deploy/llm/mlm/deploy_triton.py

@@ -209,13 +209,6 @@ def get_args(argv):
         default="gpt",
         help="Type of model to load",
     )
-    parser.add_argument(
-        "-mf",
-        "--model_format",
-        type=str,
-        default="nemo",
-        help="Format of model to load",
-    )
     parser.add_argument(
         "-mb",
         "--micro_batch_size",