easydel.kernels.gpu_ops._deprecated.pallas_gqa_flash_attention_2#
- easydel.kernels.gpu_ops._deprecated.pallas_gqa_flash_attention_2.pallas_gqa_flash_attention2_gpu(query: Array, key: Array, value: Array, bias: Optional[Array] = None, dtype: dtype = None, BLOCK_M: int = 128, BLOCK_N: int = 128, softmax_scale: float = None, inference_mode: bool = False, interpret=False, debug=False)#