Fix missing propagation of RoPE scaling and interpolation arguments in model builders

gpt_builders.py

@@ -106,6 +106,8 @@ def gpt_builder(args, pre_process, post_process, vp_stage=None, config=None, pg_
             rotary_percent=args.rotary_percent,
             rotary_base=args.rotary_base,
             rope_scaling=args.use_rope_scaling,
+            rope_scaling_factor=args.rope_scaling_factor,
+            seq_len_interpolation_factor=args.seq_len_interpolation_factor,
             mtp_block_spec=mtp_block_spec,
             vp_stage=vp_stage,
             pg_collection=pg_collection,

mamba_builders.py

@@ -38,6 +38,7 @@ def mamba_builder(args, pre_process, post_process, vp_stage=None, config=None, p
         position_embedding_type=args.position_embedding_type,
         rotary_percent=args.rotary_percent,
         rotary_base=args.rotary_base,
+        seq_len_interpolation_factor=args.seq_len_interpolation_factor,
         pg_collection=pg_collection,
     )
 

pretrain_bert.py

@@ -64,6 +64,7 @@ def model_provider(pre_process=True, post_process=True, vp_stage=None, config=No
             add_binary_head=args.bert_binary_head,
             share_embeddings_and_output_weights=not args.untie_embeddings_and_output_weights,
             parallel_output=True,
+            seq_len_interpolation_factor=args.seq_len_interpolation_factor,
             pre_process=pre_process,
             post_process=post_process,
             vp_stage=vp_stage)

pretrain_t5.py

@@ -140,6 +140,7 @@ def model_provider(
             rotary_percent=args.rotary_percent,
             relative_attention_num_buckets=args.relative_attention_num_buckets,
             relative_attention_max_distance=args.relative_attention_max_distance,
+            seq_len_interpolation_factor=args.seq_len_interpolation_factor,
             add_encoder=add_encoder,
             add_decoder=add_decoder,
         )