From 2e8265ae1764d6288aab0e2df641909072e2d58e Mon Sep 17 00:00:00 2001 From: Keiichi Tabata Date: Sun, 6 Aug 2023 15:34:05 +0900 Subject: [PATCH] convert.py : add missing abstract methods for quantized data (#2491) --- convert.py | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/convert.py b/convert.py index ab6a4e10e..f3bf17980 100644 --- a/convert.py +++ b/convert.py @@ -465,6 +465,13 @@ class GGMLQuantizedTensor(Tensor): def permute(self, n_head: int, n_kv_head: Optional[int] = None) -> 'GGMLQuantizedTensor': return GGMLQuantizedTensor(permute(self.ndarray, n_head, n_kv_head), self.shape, self.data_type) + def permute_part(self, n_part: int, n_head: int) -> 'UnquantizedTensor': + r = self.ndarray.shape[0] // 3 + return UnquantizedTensor(permute(self.ndarray[r * n_part : r * n_part + r, ...], n_head)) + + def part(self, n_part: int) -> 'UnquantizedTensor': + r = self.ndarray.shape[0] // 3 + return UnquantizedTensor(self.ndarray[r * n_part : r * n_part + r, ...]) GGMLCompatibleTensor = Union[UnquantizedTensor, GGMLQuantizedTensor]