In the repository directory, run
git submodule init
git submodule update
Then go to the k_means_kernels directory and build the kernel according to https://github.com/Zhichenzzz/k_means_kernels/blob/main/readme.md.
Should Only Be Used in Dataset Generation
- dataset: str,
- model: str,
- data_num: int,
- max_len: int,
- attn: str,
- is_cache: bool,
- seed: int,
- max_token: int,
- is_tmp: bool,
Base Class
- model: str,
- dataset: str,
- data_idx: int,
- attn: str,
- layer_ids: List[int],
- head_ids: List[int],
- round_ids: List[int],
Used when spliting into page is needed
- page_size: int,
- chunk_size: int,
Used in testing clustering algorithm
- cluster_algo: str,
- n_clusters: int,