As shown in Table 3 in the paper, we compare the generation efficiency of several generative NER models. When running on a single GeForce RTX 3090, the DiffusionNER model, using a denoising timestep of 1, demonstrated an impressive entity generation rate of ~80 sentences per second.
As shown in Table 3 in the paper, we compare the generation efficiency of several generative NER models. When running on a single GeForce RTX 3090, the DiffusionNER model, using a denoising timestep of 1, demonstrated an impressive entity generation rate of ~80 sentences per second.