[SUPPORT] Can't load quantised Gpt-oss-20b. (MxFP4)

[nandhakumarsuriya]

Gpt-oss finetuning :
While loading gpt-oss-20b model quantised on H200 GPU, with the below config:

quantization_config = Mxfp4Config(dequantize=False)
model_kwargs = dict(
    attn_implementation="eager",
    torch_dtype='auto',
    quantization_config=quantization_config,
    use_cache=False,
    device_map="auto",
)

Error:
MXFP4 quantization requires triton >= 3.4.0 and kernels installed, we will default to dequantizing the model to bf16

Note:
I have also installed triton.

Still why can't i load model with Mxfp4 quantized version.