fix vllm comments

Signed-off-by: Hung-Yueh Chiang <hungyuehc@nvidia.com>

Author: hychiang-git

CHANGELOG.rst

@@ -27,7 +27,7 @@ Changelog
 
 **New Features**
 
-- Add NVFP4 W4A16 weight-only quantization (``w4a16_nvfp4``): FP4 weights with group_size=16, BF16 activations, no calibration forward pass required. Use ``mtq.W4A16_NVFP4_CFG`` or ``--qformat w4a16_nvfp4`` in ``hf_ptq.py``. Exported checkpoints can be served on vLLM after conversion to compressed-tensors format.
+- Add NVFP4 W4A16 weight-only quantization (``w4a16_nvfp4``): FP4 weights with group_size=16, BF16 activations, no calibration forward pass required. Use ``mtq.W4A16_NVFP4_CFG`` or ``--qformat w4a16_nvfp4`` in ``hf_ptq.py``. vLLM deployment support is in progress.
 - Support full Transformer Engine spec for Minitron pruning (``mcore_minitron``). Now we no longer need to use custom ModelOpt spec. Note that this does not affect the usage of the pruning workflow but makes pruning slightly faster and may result in slightly different pruned model because of different kernel and numerics.
 - Add end-to-end tutorial for Minitron pruning + distillation + quantization + evaluation + vLLM deployment for Nemotron-Nano-9B-v2 → Pruned 7B along with data blend preparation steps (and ablation study). See `examples/pruning/minitron/README.md <https://github.com/NVIDIA/Model-Optimizer/tree/main/examples/pruning/minitron/>`_ for details.
 - Add Puzzletron - a new algorithm for heterogeneous pruning of LLM and VLM models. See `examples/puzzletron/README.md <https://github.com/NVIDIA/Model-Optimizer/tree/main/examples/puzzletron>`_ for more details.

examples/llm_ptq/hf_ptq.py

@@ -789,7 +789,7 @@ def export_quantized(
                 if args.qformat == "w4a16_nvfp4":
                     warnings.warn(
                         "TensorRT-LLM and SGLang do not support this format. "
-                        "To serve on vLLM, convert the NVFP4 W4A16 checkpoint to compressed-tensors format."
+                        "vLLM deployment support is in progress."
                     )
 
         # Restore default padding and export the tokenizer as well.

modelopt/torch/export/quant_utils.py

@@ -66,10 +66,10 @@
     QUANTIZATION_NVFP4,
     QUANTIZATION_NVFP4_AWQ,
     QUANTIZATION_NVFP4_SVDQUANT,
-    QUANTIZATION_W4A16_NVFP4,
     QUANTIZATION_W4A8_AWQ,
     QUANTIZATION_W4A8_MXFP4_FP8,
     QUANTIZATION_W4A8_NVFP4_FP8,
+    QUANTIZATION_W4A16_NVFP4,
 )
 
 logger = logging.getLogger(__name__)

modelopt/torch/export/unified_export_hf.py

@@ -80,9 +80,9 @@
     QUANTIZATION_NVFP4,
     QUANTIZATION_NVFP4_AWQ,
     QUANTIZATION_NVFP4_SVDQUANT,
-    QUANTIZATION_W4A16_NVFP4,
     QUANTIZATION_W4A8_AWQ,
     QUANTIZATION_W4A8_NVFP4_FP8,
+    QUANTIZATION_W4A16_NVFP4,
 )
 from .model_utils import get_language_model_from_vl, is_multimodal_model
 from .moe_utils import _export_fused_experts