hjxdl 0.0.13__py3-none-any.whl → 0.0.15__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- hdl/_version.py +2 -2
- hdl/utils/llm/embs.py +27 -13
- {hjxdl-0.0.13.dist-info → hjxdl-0.0.15.dist-info}/METADATA +1 -1
- {hjxdl-0.0.13.dist-info → hjxdl-0.0.15.dist-info}/RECORD +6 -6
- {hjxdl-0.0.13.dist-info → hjxdl-0.0.15.dist-info}/WHEEL +0 -0
- {hjxdl-0.0.13.dist-info → hjxdl-0.0.15.dist-info}/top_level.txt +0 -0
hdl/_version.py
CHANGED
hdl/utils/llm/embs.py
CHANGED
@@ -1,23 +1,33 @@
|
|
1
|
-
|
2
|
-
|
3
|
-
|
4
|
-
class FlagEmbedder():
|
1
|
+
class HFEmbedder():
|
5
2
|
def __init__(
|
6
3
|
self,
|
4
|
+
emb_name: str = "bge",
|
7
5
|
emb_dir: str = None
|
8
6
|
) -> None:
|
9
|
-
"""
|
7
|
+
"""Initializes the object with the specified embedding name and directory.
|
10
8
|
|
11
9
|
Args:
|
12
|
-
|
13
|
-
|
10
|
+
emb_name (str): The name of the embedding. Defaults to "bge".
|
11
|
+
emb_dir (str): The directory path for the embedding model.
|
12
|
+
|
14
13
|
Returns:
|
15
14
|
None
|
16
15
|
"""
|
17
|
-
self.
|
18
|
-
|
19
|
-
|
20
|
-
)
|
16
|
+
self.emb_name = emb_name
|
17
|
+
self.emb_dir = emb_dir
|
18
|
+
|
19
|
+
if "bge" in emb_name.lower():
|
20
|
+
from FlagEmbedding import BGEM3FlagModel
|
21
|
+
self.model = BGEM3FlagModel(
|
22
|
+
emb_dir,
|
23
|
+
use_fp16=True
|
24
|
+
)
|
25
|
+
elif "bce" in emb_name.lower():
|
26
|
+
from BCEmbedding import EmbeddingModel
|
27
|
+
self.model = EmbeddingModel(
|
28
|
+
model_name_or_path=emb_dir,
|
29
|
+
use_fp16=True
|
30
|
+
)
|
21
31
|
|
22
32
|
def encode(
|
23
33
|
self,
|
@@ -31,12 +41,16 @@ class FlagEmbedder():
|
|
31
41
|
Returns:
|
32
42
|
numpy.ndarray: Encoded representation of the input sentences.
|
33
43
|
"""
|
44
|
+
if isinstance(sentences, str):
|
45
|
+
sentences = [sentences]
|
34
46
|
output = self.model.encode(
|
35
47
|
sentences,
|
36
48
|
return_dense=True,
|
37
49
|
return_sparse=True,
|
38
50
|
return_colbert_vecs=False
|
39
51
|
)
|
52
|
+
if "bge" in self.emb_name.lower():
|
53
|
+
return output["dense_vecs"]
|
40
54
|
return output
|
41
55
|
|
42
56
|
def sim(
|
@@ -55,5 +69,5 @@ class FlagEmbedder():
|
|
55
69
|
"""
|
56
70
|
output_1 = self.encode(sentences_1)
|
57
71
|
output_2 = self.encode(sentences_2)
|
58
|
-
similarity = output_1
|
59
|
-
return similarity
|
72
|
+
similarity = output_1 @ output_2.T
|
73
|
+
return similarity
|
@@ -1,5 +1,5 @@
|
|
1
1
|
hdl/__init__.py,sha256=5sZZNySv08wwfzJcSDssGTqUn9wlmDsR6R4XB8J8mFM,70
|
2
|
-
hdl/_version.py,sha256=
|
2
|
+
hdl/_version.py,sha256=pzShTmESM-5FMm5QD9BQxtLcs8dS3tIY17WJZlF9dSQ,413
|
3
3
|
hdl/args/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
4
4
|
hdl/args/loss_args.py,sha256=s7YzSdd7IjD24rZvvOrxLLFqMZQb9YylxKeyelSdrTk,70
|
5
5
|
hdl/controllers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
@@ -85,10 +85,10 @@ hdl/utils/general/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU
|
|
85
85
|
hdl/utils/general/glob.py,sha256=8-RCnt6L297wMIfn34ZAMCsGCZUjHG3MGglGZI1cX0g,491
|
86
86
|
hdl/utils/llm/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
87
87
|
hdl/utils/llm/chat.py,sha256=NEbWdrPpQqS5ArU7ghA6xuam8wwvWXk3LL_CiNEEuIg,7055
|
88
|
-
hdl/utils/llm/embs.py,sha256=
|
88
|
+
hdl/utils/llm/embs.py,sha256=yCFtc25gUFas6kwgOGBFydeaHNyQMq5y1Chxl8TNEUQ,2190
|
89
89
|
hdl/utils/schedulers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
90
90
|
hdl/utils/schedulers/norm_lr.py,sha256=bDwCmdEK-WkgxQMFBiMuchv8Mm7C0-GZJ6usm-PQk14,4461
|
91
|
-
hjxdl-0.0.
|
92
|
-
hjxdl-0.0.
|
93
|
-
hjxdl-0.0.
|
94
|
-
hjxdl-0.0.
|
91
|
+
hjxdl-0.0.15.dist-info/METADATA,sha256=uVY_O0levGtYTHXWtCIstsHzulDmM4JLwQ7Xu6uZ5sg,543
|
92
|
+
hjxdl-0.0.15.dist-info/WHEEL,sha256=Z4pYXqR_rTB7OWNDYFOm1qRk0RX6GFP2o8LgvP453Hk,91
|
93
|
+
hjxdl-0.0.15.dist-info/top_level.txt,sha256=-kxwTM5JPhylp06z3zAVO3w6_h7wtBfBo2zgM6YZoTk,4
|
94
|
+
hjxdl-0.0.15.dist-info/RECORD,,
|
File without changes
|
File without changes
|