Update flag_models.py

545999961 · web-flow · commit 566781028787 · 2024-07-29T23:32:07.000+08:00
diff --git a/FlagEmbedding/flag_models.py b/FlagEmbedding/flag_models.py
@@ -121,7 +121,7 @@ def encode_queries(self, queries: Union[List[str], str],
                 return_tensors=None,
                 add_special_tokens=False
             )
-            new_max_length = (len(prefix_ids) + len(suffix_ids) + max_length) // 8 * 8 + 8
+            new_max_length = (len(prefix_ids) + len(suffix_ids) + max_length + 8) // 8 * 8 + 8
             sentences_batch = self.tokenizer.batch_decode(inputs['input_ids'])
             for i in range(len(sentences_batch)):
                 sentences_batch[i] = self.prefix + sentences_batch[i] + self.suffix

Original file line number	Diff line number	Diff line change
`@@ -121,7 +121,7 @@ def encode_queries(self, queries: Union[List[str], str],`
`121`	`121`	`return_tensors=None,`
`122`	`122`	`add_special_tokens=False`
`123`	`123`	`)`
`124`		`- new_max_length = (len(prefix_ids) + len(suffix_ids) + max_length) // 8 * 8 + 8`
	`124`	`+ new_max_length = (len(prefix_ids) + len(suffix_ids) + max_length + 8) // 8 * 8 + 8`
`125`	`125`	`sentences_batch = self.tokenizer.batch_decode(inputs['input_ids'])`
`126`	`126`	`for i in range(len(sentences_batch)):`
`127`	`127`	`sentences_batch[i] = self.prefix + sentences_batch[i] + self.suffix`