A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
[Model] rename NemotronH_Nano_VL -> NemotronH_Nano_VL_V2 (#25708)
Signed-off-by: Tomer Asida <57313761+tomeras91@users.noreply.github.com>
tomeras91 committed
Sep 26, 2025 at 02:10 UTC
57329a8c013ca9e5d575faad3f04436f2eabad15
3 files changed
+6
-6
tests/models/registry.py
+1
-1
@@ -531,7 +531,7 @@ _MULTIMODAL_EXAMPLE_MODELS = {
531
trust_remote_code=True),
532
"Llama_Nemotron_Nano_VL" : _HfExamplesInfo("nvidia/Llama-3.1-Nemotron-Nano-VL-8B-V1", # noqa: E501
533
trust_remote_code=True),
534
- "NemotronH_Nano_VL": _HfExamplesInfo("nano_vl_dummy",
534
+ "NemotronH_Nano_VL_V2": _HfExamplesInfo("nano_vl_dummy",
535
is_available_online=False,
536
trust_remote_code=True),
537
"Ovis": _HfExamplesInfo("AIDC-AI/Ovis2-1B", trust_remote_code=True,
vllm/model_executor/models/nano_nemotron_vl.py
+4
-4
@@ -869,8 +869,8 @@ class NanoNemotronVLDummyInputsBuilder(
869
info=NanoNemotronVLProcessingInfo,
870
dummy_inputs=NanoNemotronVLDummyInputsBuilder,
871
)
872
-class NemotronH_Nano_VL(nn.Module, HasInnerState, IsHybrid,
873
- SupportsMultiModal):
872
+class NemotronH_Nano_VL_V2(nn.Module, HasInnerState, IsHybrid,
873
+ SupportsMultiModal):
874
875
@classmethod
876
def get_placeholder_str(cls, modality: str, i: int) -> Optional[str]:
@@ -1249,7 +1249,7 @@ class NemotronH_Nano_VL(nn.Module, HasInnerState, IsHybrid,
1249
1250
try:
1251
print("=" * 100)
1252
- print("NemotronH_Nano_VL Model Architecture")
1252
+ print("NemotronH_Nano_VL_V2 Model Architecture")
1253
print("=" * 100)
1254
1255
total_params = 0
@@ -1333,7 +1333,7 @@ class NemotronH_Nano_VL(nn.Module, HasInnerState, IsHybrid,
1333
component_info[component]["size"] += param.numel()
1334
1335
return {
1336
- "model_name": "NemotronH_Nano_VL",
1336
+ "model_name": "NemotronH_Nano_VL_V2",
1337
"total_parameters": total_params,
1338
"memory_estimate_mb": total_params * 2 / (1024**2), # bfloat16
1339
"components": component_info,
vllm/model_executor/models/registry.py
+1
-1
@@ -231,7 +231,7 @@ _MULTIMODAL_MODELS = {
231
"GraniteSpeechForConditionalGeneration": ("granite_speech", "GraniteSpeechForConditionalGeneration"), # noqa: E501
232
"H2OVLChatModel": ("h2ovl", "H2OVLChatModel"),
233
"InternVLChatModel": ("internvl", "InternVLChatModel"),
234
- "NemotronH_Nano_VL": ("nano_nemotron_vl", "NemotronH_Nano_VL"),
234
+ "NemotronH_Nano_VL_V2": ("nano_nemotron_vl", "NemotronH_Nano_VL_V2"),
235
"InternS1ForConditionalGeneration": ("interns1", "InternS1ForConditionalGeneration"), # noqa: E501
236
"InternVLForConditionalGeneration": ("interns1", "InternS1ForConditionalGeneration"), # noqa: E501
237
"Idefics3ForConditionalGeneration":("idefics3","Idefics3ForConditionalGeneration"),