Convert: Make NVFP4 and MXFP4 HF conversions say NVFP4/MXFP4 instead of BF16

[michaelw9999]

When converting an NVFP4 or MXFP4 HF model to GGUF, the default behavior is to keep the same name if not specified, so the converted models still are named BF16:

Before:

| model                          |     
| ------------------------------ | 
| qwen35 0.8B BF16               |

After:

| model                          |     
| ------------------------------ | 
| qwen35 0.8B NVFP4              |

This script edit identifies the file type properly for both NVFP4 and MXFP4 and adds the missing MOSTLY_NVFP4 and missing MOSTLY_MXFP4_MOE from the table in the py constants file.

[CISC]

This simplifies things while also fixing MXFP4.

Add quant_method = (self.hparams.get("quantization_config") or {}).get("quant_method") here

llama.cpp/convert_hf_to_gguf.py

Line 714 in 5d41b4c

quant_algo = (self.hparams.get("quantization_config") or {}).get("quant_algo")

and add self._is_mxfp4 = quant_method == "mxfp4" here

llama.cpp/convert_hf_to_gguf.py

Line 730 in 5d41b4c

self._is_nvfp4 = quant_algo == "NVFP4"

then update this

llama.cpp/convert_hf_to_gguf.py

Lines 11126 to 11131 in 5d41b4c

	# TODO: remove once MXFP4 is supported more generally
	def dequant_model(self):
	quant_config = self.hparams.get("quantization_config")
	if quant_config is not None and quant_config.get("quant_method") == "mxfp4":
	return
	return super().dequant_model()

[ggerganov]

Should these be case insensitive?

[CISC]

Not as far as I've been able to tell, quant_method seems to be always lower-case, while quant_algo (at least for NVFP4) is always uppper-case.