Pavelrst

Here is a simple example: ``` data = np.array([[1,1]] * 500) clusterer = hdbscan.HDBSCAN(min_cluster_size=5, cluster_selection_method='eom').fit(data) clusterer.labels_ ``` result: ``` array([ 1, -1, -1, 0, 1, 1, 1, 1, 1, 1,...

@jonathlela I have the same issue with optimized t5-base. You mentioned that disabling "replace_layer_norm_rms" fixes the issue. How replace_layer_norm_rms can be disabled? I've tried commenting out this line and re-running...