Questioning bartowski's MXFP4 claim for DeepSeek-V4-Flash

A Reddit user questions why the Hugging Face repository by bartowski claims that the DeepSeek-V4-Flash model is in MXFP4 format. The original model page from deepseek-ai lists tensor types as BF16, I64, F32, F8_E8M0, F8_E4M3, and I8, with no mention of MXFP4.

• bartowski's GGUF repository states the model is provided exclusively in MXFP4 format.
• The original deepseek-ai repository lists BF16, I64, F32, F8_E8M0, F8_E4M3, and I8 as tensor types.
• The user asks if they are missing something regarding this discrepancy.

The post seeks clarification on whether the MXFP4 format is a valid conversion or an error in the repository description.