From: Keiichi Tabata <redacted>
Date: Sun, 6 Aug 2023 06:34:05 +0000 (+0900)
Subject: convert.py : add missing abstract methods for quantized data (#2491)
X-Git-Tag: gguf-v0.4.0~352
X-Git-Url: https://git.djapps.eu/?a=commitdiff_plain;h=2e8265ae1764d6288aab0e2df641909072e2d58e;p=pkg%2Fggml%2Fsources%2Fllama.cpp

convert.py : add missing abstract methods for quantized data (#2491)
---

diff --git a/convert.py b/convert.py
index ab6a4e10..f3bf1798 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]