ai-edge-torch-nightly 0.6.0.dev20250619__py3-none-any.whl → 0.6.0.dev20250620__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.
@@ -24,26 +24,27 @@ import torch.nn.functional as F
24
24
 
25
25
 
26
26
  TENSOR_NAMES = loading_utils.ModelLoader.TensorNames(
27
- ff_up_proj="vision_tower.vision_model.encoder.layers.{}.mlp.fc1",
28
- ff_down_proj="vision_tower.vision_model.encoder.layers.{}.mlp.fc2",
27
+ ff_up_proj="siglip_vision_model.encoder_blocks.{}.mlp.fc1",
28
+ ff_down_proj="siglip_vision_model.encoder_blocks.{}.mlp.fc2",
29
29
  attn_query_proj=(
30
- "vision_tower.vision_model.encoder.layers.{}.self_attn.q_proj"
30
+ "siglip_vision_model.encoder_blocks.{}.self_attn.q_proj"
31
31
  ),
32
32
  attn_key_proj=(
33
- "vision_tower.vision_model.encoder.layers.{}.self_attn.k_proj"
33
+ "siglip_vision_model.encoder_blocks.{}.self_attn.k_proj"
34
34
  ),
35
35
  attn_value_proj=(
36
- "vision_tower.vision_model.encoder.layers.{}.self_attn.v_proj"
36
+ "siglip_vision_model.encoder_blocks.{}.self_attn.v_proj"
37
37
  ),
38
38
  attn_output_proj=(
39
- "vision_tower.vision_model.encoder.layers.{}.self_attn.out_proj"
39
+ "siglip_vision_model.encoder_blocks.{}.self_attn.o_proj"
40
40
  ),
41
- pre_attn_norm="vision_tower.vision_model.encoder.layers.{}.layer_norm1",
42
- embedding="vision_tower.vision_model.embeddings.patch_embedding",
41
+ pre_attn_norm="siglip_vision_model.encoder_blocks.{}.layer_norm1",
42
+ pre_ff_norm="siglip_vision_model.encoder_blocks.{}.layer_norm2",
43
+ embedding="siglip_vision_model.patch_embedding",
43
44
  embedding_position=(
44
- "vision_tower.vision_model.embeddings.position_embedding.weight"
45
+ "siglip_vision_model.position_embedding.weight"
45
46
  ),
46
- final_norm="vision_tower.vision_model.post_layernorm",
47
+ final_norm="siglip_vision_model.final_norm",
47
48
  )
48
49
 
49
50
 
@@ -61,6 +61,7 @@ def build_causal_mask_cache(
61
61
  size: int,
62
62
  dtype: torch.dtype = torch.float32,
63
63
  device: torch.device = None,
64
+ mask_value: float = float('-inf'),
64
65
  ) -> torch.Tensor:
65
66
  """Build a cache for causal attention mask.
66
67
 
@@ -70,6 +71,8 @@ def build_causal_mask_cache(
70
71
  torch.float32.
71
72
  device (torch.device, optional): Output tensor's data type. Defaults to
72
73
  None in which case "cpu" is used.
74
+ mask_value (float, optional): The value to set the mask to. Defaults to
75
+ float('-inf').
73
76
 
74
77
  Returns:
75
78
  torch.Tensor: Causal attention mask.
@@ -77,7 +80,7 @@ def build_causal_mask_cache(
77
80
 
78
81
  if device is None:
79
82
  device = torch.device('cpu')
80
- mask = torch.full((size, size), float('-inf'), dtype=dtype, device=device)
83
+ mask = torch.full((size, size), mask_value, dtype=dtype, device=device)
81
84
  return torch.triu(mask, diagonal=1).unsqueeze(0).unsqueeze(0)
82
85
 
83
86
 
@@ -86,6 +89,7 @@ def build_sliding_window_mask_cache(
86
89
  window_size: int,
87
90
  dtype: torch.dtype = torch.float32,
88
91
  device: torch.device = None,
92
+ mask_value: float = float('-inf'),
89
93
  ) -> torch.Tensor:
90
94
  """Build a cache for a sliding window mask.
91
95
 
@@ -96,18 +100,20 @@ def build_sliding_window_mask_cache(
96
100
  torch.float32.
97
101
  device (torch.device, optional): Output tensor's data type. Defaults to
98
102
  None in which case "cpu" is used.
103
+ mask_value (float, optional): The value to set the mask to. Defaults to
104
+ float('-inf').
99
105
 
100
106
  Returns:
101
107
  torch.Tensor: Causal attention mask.
102
108
  """
103
109
 
104
- mask = build_causal_mask_cache(size, dtype, device)
110
+ mask = build_causal_mask_cache(size, dtype, device, mask_value)
105
111
  all_ones = torch.ones_like(mask)
106
112
  window_size = min(size, window_size)
107
113
  sliding_mask = torch.triu(all_ones, -1 * window_size + 1) * torch.tril(
108
114
  all_ones, window_size - 1
109
115
  )
110
- return torch.where(sliding_mask == 1, mask, float('-inf'))
116
+ return torch.where(sliding_mask == 1, mask, mask_value)
111
117
 
112
118
 
113
119
  def relative_position_bucket(
ai_edge_torch/version.py CHANGED
@@ -15,4 +15,4 @@
15
15
 
16
16
  # The next version of ai-edge-torch.
17
17
  # The minor version code should be bumped after every release.
18
- __version__ = "0.6.0.dev20250619"
18
+ __version__ = "0.6.0.dev20250620"
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: ai-edge-torch-nightly
3
- Version: 0.6.0.dev20250619
3
+ Version: 0.6.0.dev20250620
4
4
  Summary: Supporting PyTorch models with the Google AI Edge TFLite runtime.
5
5
  Home-page: https://github.com/google-ai-edge/ai-edge-torch
6
6
  Keywords: On-Device ML,AI,Google,TFLite,PyTorch,LLMs,GenAI
@@ -2,7 +2,7 @@ ai_edge_torch/__init__.py,sha256=lemyLCNoGYRnJsmDuGZu7qOqLbLqG6CGDFtu3ue1syU,129
2
2
  ai_edge_torch/_config.py,sha256=AiqhbcheF7j_ozIGDLC89k1we95aVgFDa-tR6h7UI0s,2529
3
3
  ai_edge_torch/conftest.py,sha256=r0GTrhMRhlmOGrrkvumHN8hkmyug6WvF60vWq8wRIBI,758
4
4
  ai_edge_torch/model.py,sha256=A7loFu8jE9CsXsfMmHYZ-KDFJiaD8Kkqwm_9d3IVzk0,5638
5
- ai_edge_torch/version.py,sha256=aCO6sA_1IPQGd5f8Ya-ce4ZKJE1EEt2BkypXJLQ3qvI,806
5
+ ai_edge_torch/version.py,sha256=L-kCkN9vMaBYCuarY-Y8kGgttAEHZdyPWaupqKVMLiA,806
6
6
  ai_edge_torch/_convert/__init__.py,sha256=hHLluseD2R0Hh4W6XZRIXY_dRQeYudjsrKGf6LZz65g,671
7
7
  ai_edge_torch/_convert/conversion.py,sha256=iQk3R-pLq4c1nfLqPB4xTRj78gghxPGzJCJtILLdg5o,6123
8
8
  ai_edge_torch/_convert/conversion_utils.py,sha256=Sr8qXVcTwc-ZnZmK7yxVrIOOp1S_vNrwzC0zUvLTI2o,2160
@@ -74,7 +74,7 @@ ai_edge_torch/generative/examples/gemma3/__init__.py,sha256=JaAnrFoXTl3RJX97Xspk
74
74
  ai_edge_torch/generative/examples/gemma3/convert_gemma3_to_tflite.py,sha256=UEDNN3JmI31WfE2pvacxeJpqumKK86L2dEus3yTURaY,2114
75
75
  ai_edge_torch/generative/examples/gemma3/decoder.py,sha256=1UVv9SFFg5degX3wf-Fefx7nor1AzJj2NWBVuo8bRnM,15540
76
76
  ai_edge_torch/generative/examples/gemma3/gemma3.py,sha256=fFMyIS8si3GdwW8EsdhYk1OKyg_27xDv1HTQ2Gv4N8E,6616
77
- ai_edge_torch/generative/examples/gemma3/image_encoder.py,sha256=tUOI99kdny33qcDM7-z0R6F-1aU1lZ24kG5zeLVdwow,5129
77
+ ai_edge_torch/generative/examples/gemma3/image_encoder.py,sha256=OCMIAQfNmPR4uQUAtlYL6j4xkG0dw2Ays4-lnThcWqQ,5110
78
78
  ai_edge_torch/generative/examples/gemma3/verify_gemma3.py,sha256=v8oNXFICmVOtQxfO7IhZ8GnbvotEkDi9lzYHjoQyOso,2464
79
79
  ai_edge_torch/generative/examples/gemma3/verify_util.py,sha256=5OmUwz38kVHYLA-v8U8evvDN9da2WioZtGo-XK6yq1o,10067
80
80
  ai_edge_torch/generative/examples/hammer/__init__.py,sha256=hHLluseD2R0Hh4W6XZRIXY_dRQeYudjsrKGf6LZz65g,671
@@ -171,7 +171,7 @@ ai_edge_torch/generative/fx_passes/remove_sdpa_zero_mask_pass.py,sha256=myGjal5A
171
171
  ai_edge_torch/generative/layers/__init__.py,sha256=hHLluseD2R0Hh4W6XZRIXY_dRQeYudjsrKGf6LZz65g,671
172
172
  ai_edge_torch/generative/layers/attention.py,sha256=RaXENRRQo1MsLdt3U8h3kYTCmd6imHQ-aCXtmPXCh_o,13911
173
173
  ai_edge_torch/generative/layers/attention_test.py,sha256=9v8v96TLyFPdqxEylU1JOAeRFAp2s0YoDHZN83SFJJA,4764
174
- ai_edge_torch/generative/layers/attention_utils.py,sha256=zBVwlBUTs-nStIKCZG0ks5ra7tsqc9ShfakFJKH5rds,7344
174
+ ai_edge_torch/generative/layers/attention_utils.py,sha256=2qfg7Tzk9ikKph5w3geOHC1I6EyOCdDsWXMr7F7IOZM,7630
175
175
  ai_edge_torch/generative/layers/attention_utils_test.py,sha256=22gQ1gcRPkwqFG3_p82GZfRKVE3udEssSy58wNOqv0w,2431
176
176
  ai_edge_torch/generative/layers/builder.py,sha256=2bUgkyowDkDznkF8XaHyZs4nowHr1QEHYLM7pMaFmIk,4921
177
177
  ai_edge_torch/generative/layers/einsum.py,sha256=EsZSWNVWUs0-1plp4TBnhP4ZhaRDBa2VlDO6hWpUAqU,1288
@@ -268,8 +268,8 @@ ai_edge_torch/testing/__init__.py,sha256=_yGgvnBZWb7T3IN3mc4x1sS4vM96HZwM8pwIcPG
268
268
  ai_edge_torch/testing/export.py,sha256=k5mGDGzwc23Z4zaIVDs8CNh-oOt64gsf9MS9NjhbPy4,3293
269
269
  ai_edge_torch/testing/model_coverage/__init__.py,sha256=5P8J6Zk5YYtDvTBucFvB9NGSRI7Gw_24WnrbhXgycEE,765
270
270
  ai_edge_torch/testing/model_coverage/model_coverage.py,sha256=UPB448aMDUyC0HNYVqio2rcJPnDN0tBQMP08J6vPYew,4718
271
- ai_edge_torch_nightly-0.6.0.dev20250619.dist-info/LICENSE,sha256=z8d0m5b2O9McPEK1xHG_dWgUBT6EfBDz6wA0F7xSPTA,11358
272
- ai_edge_torch_nightly-0.6.0.dev20250619.dist-info/METADATA,sha256=_nUnboHwt2qZ0ejpyggXZf5q5iuNSNUUboXq6e8uQGw,2074
273
- ai_edge_torch_nightly-0.6.0.dev20250619.dist-info/WHEEL,sha256=tZoeGjtWxWRfdplE7E3d45VPlLNQnvbKiYnx7gwAy8A,92
274
- ai_edge_torch_nightly-0.6.0.dev20250619.dist-info/top_level.txt,sha256=5KXRaF2hwkApYxf7Y8y_tVb9aulGTlbOoNdbx1aKRkE,14
275
- ai_edge_torch_nightly-0.6.0.dev20250619.dist-info/RECORD,,
271
+ ai_edge_torch_nightly-0.6.0.dev20250620.dist-info/LICENSE,sha256=z8d0m5b2O9McPEK1xHG_dWgUBT6EfBDz6wA0F7xSPTA,11358
272
+ ai_edge_torch_nightly-0.6.0.dev20250620.dist-info/METADATA,sha256=W4cZLDBaywmznVjn7haIlLen5cKvXF7VYVIapnx4h0E,2074
273
+ ai_edge_torch_nightly-0.6.0.dev20250620.dist-info/WHEEL,sha256=tZoeGjtWxWRfdplE7E3d45VPlLNQnvbKiYnx7gwAy8A,92
274
+ ai_edge_torch_nightly-0.6.0.dev20250620.dist-info/top_level.txt,sha256=5KXRaF2hwkApYxf7Y8y_tVb9aulGTlbOoNdbx1aKRkE,14
275
+ ai_edge_torch_nightly-0.6.0.dev20250620.dist-info/RECORD,,