ai-edge-torch-nightly 0.6.0.dev20250528__py3-none-any.whl → 0.6.0.dev20250530__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.
- ai_edge_torch/generative/layers/model_config.py +22 -2
- ai_edge_torch/generative/utilities/test_utils.py +29 -0
- ai_edge_torch/model.py +2 -1
- ai_edge_torch/version.py +1 -1
- {ai_edge_torch_nightly-0.6.0.dev20250528.dist-info → ai_edge_torch_nightly-0.6.0.dev20250530.dist-info}/METADATA +1 -1
- {ai_edge_torch_nightly-0.6.0.dev20250528.dist-info → ai_edge_torch_nightly-0.6.0.dev20250530.dist-info}/RECORD +9 -8
- {ai_edge_torch_nightly-0.6.0.dev20250528.dist-info → ai_edge_torch_nightly-0.6.0.dev20250530.dist-info}/LICENSE +0 -0
- {ai_edge_torch_nightly-0.6.0.dev20250528.dist-info → ai_edge_torch_nightly-0.6.0.dev20250530.dist-info}/WHEEL +0 -0
- {ai_edge_torch_nightly-0.6.0.dev20250528.dist-info → ai_edge_torch_nightly-0.6.0.dev20250530.dist-info}/top_level.txt +0 -0
@@ -134,8 +134,6 @@ class AttentionConfig:
|
|
134
134
|
value_norm_config: NormalizationConfig = dataclasses.field(
|
135
135
|
default_factory=NormalizationConfig
|
136
136
|
)
|
137
|
-
# Whether the KV cache is shared with the previous attention block.
|
138
|
-
kv_shared: bool = False
|
139
137
|
relative_attention_num_buckets: int = 0
|
140
138
|
relative_attention_max_distance: int = 0
|
141
139
|
# Softcap on the output logits.
|
@@ -268,6 +266,12 @@ class ModelConfig:
|
|
268
266
|
# export.
|
269
267
|
use_mask_cache: bool = True
|
270
268
|
|
269
|
+
# An interleaved sequence of the attention types used in the model.
|
270
|
+
# E.g. [AttentionType.LOCAL_SLIDING, AttentionType.LOCAL_SLIDING,
|
271
|
+
# AttentionType.GLOBAL] means that the model has an attention pattern of 2
|
272
|
+
# local attentions followed by a global attention in a repeated pattern.
|
273
|
+
attention_patterns: Optional[Sequence[AttentionType]] = None
|
274
|
+
|
271
275
|
@property
|
272
276
|
def kv_cache_max(self) -> int:
|
273
277
|
if self.kv_cache_max_len > 0:
|
@@ -286,3 +290,19 @@ class ModelConfig:
|
|
286
290
|
@property
|
287
291
|
def causal_mask_value(self) -> float:
|
288
292
|
return self.block_config(0).attn_config.causal_mask_value
|
293
|
+
|
294
|
+
def check_if_global_attention_layer(self, layer_idx: int) -> bool:
|
295
|
+
"""Returns True if the layer is a global attention layer."""
|
296
|
+
if self.attention_patterns is None:
|
297
|
+
# If attention_patterns is not set, we assume the model has global
|
298
|
+
# attention.
|
299
|
+
return True
|
300
|
+
assert layer_idx >= 0 and layer_idx < self.num_layers, (
|
301
|
+
"Layer index {layer_idx} is out of range for num_layers:"
|
302
|
+
f" {self.num_layers}"
|
303
|
+
)
|
304
|
+
|
305
|
+
return (
|
306
|
+
self.block_config(layer_idx).attn_config.attn_type
|
307
|
+
== AttentionType.GLOBAL
|
308
|
+
)
|
@@ -0,0 +1,29 @@
|
|
1
|
+
# Copyright 2025 The AI Edge Torch Authors.
|
2
|
+
#
|
3
|
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
4
|
+
# you may not use this file except in compliance with the License.
|
5
|
+
# You may obtain a copy of the License at
|
6
|
+
#
|
7
|
+
# http://www.apache.org/licenses/LICENSE-2.0
|
8
|
+
#
|
9
|
+
# Unless required by applicable law or agreed to in writing, software
|
10
|
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
11
|
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
12
|
+
# See the License for the specific language governing permissions and
|
13
|
+
# limitations under the License.
|
14
|
+
# ==============================================================================
|
15
|
+
"""Test utils for generative layers."""
|
16
|
+
from typing import Sequence
|
17
|
+
from ai_edge_torch.generative.layers import kv_cache as kv_utils
|
18
|
+
import torch
|
19
|
+
|
20
|
+
|
21
|
+
def initialize_kv_cache_all_zeros(
|
22
|
+
kv_shape: Sequence[int],
|
23
|
+
layout: kv_utils.KVLayout = kv_utils.KV_LAYOUT_DEFAULT,
|
24
|
+
) -> kv_utils.KVCacheEntry:
|
25
|
+
return kv_utils.KVCacheEntry(
|
26
|
+
k_cache=torch.zeros(kv_shape, dtype=torch.float32),
|
27
|
+
v_cache=torch.zeros(kv_shape, dtype=torch.float32),
|
28
|
+
kv_layout=layout,
|
29
|
+
)
|
ai_edge_torch/model.py
CHANGED
@@ -155,7 +155,8 @@ class TfLiteModel(Model):
|
|
155
155
|
Args:
|
156
156
|
path: The path to file to which the model is serialized.
|
157
157
|
"""
|
158
|
-
os.
|
158
|
+
if os.path.dirname(path):
|
159
|
+
os.makedirs(os.path.dirname(path), exist_ok=True)
|
159
160
|
with open(path, 'wb') as file_handle:
|
160
161
|
file_handle.write(self._tflite_model)
|
161
162
|
|
ai_edge_torch/version.py
CHANGED
@@ -1,6 +1,6 @@
|
|
1
1
|
Metadata-Version: 2.1
|
2
2
|
Name: ai-edge-torch-nightly
|
3
|
-
Version: 0.6.0.
|
3
|
+
Version: 0.6.0.dev20250530
|
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
|
@@ -1,8 +1,8 @@
|
|
1
1
|
ai_edge_torch/__init__.py,sha256=lemyLCNoGYRnJsmDuGZu7qOqLbLqG6CGDFtu3ue1syU,1290
|
2
2
|
ai_edge_torch/_config.py,sha256=AiqhbcheF7j_ozIGDLC89k1we95aVgFDa-tR6h7UI0s,2529
|
3
3
|
ai_edge_torch/conftest.py,sha256=r0GTrhMRhlmOGrrkvumHN8hkmyug6WvF60vWq8wRIBI,758
|
4
|
-
ai_edge_torch/model.py,sha256=
|
5
|
-
ai_edge_torch/version.py,sha256=
|
4
|
+
ai_edge_torch/model.py,sha256=A7loFu8jE9CsXsfMmHYZ-KDFJiaD8Kkqwm_9d3IVzk0,5638
|
5
|
+
ai_edge_torch/version.py,sha256=B-sOsG_3lPrDKxH_MJPNpivWVftaRufBHKPbBig2z3E,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
|
@@ -180,7 +180,7 @@ ai_edge_torch/generative/layers/feed_forward.py,sha256=_GmtHxwL068l9gh_F_WFcFk7L
|
|
180
180
|
ai_edge_torch/generative/layers/feed_forward_test.py,sha256=Y5l1eC9NgfYixHcfIfE1W4FGh7oC-9UGGyHdKS9tQKc,1880
|
181
181
|
ai_edge_torch/generative/layers/kv_cache.py,sha256=b-7shzDaKexmvQF7P3SiAmIz4ZofjYWv3m5u71GojsA,10460
|
182
182
|
ai_edge_torch/generative/layers/lora.py,sha256=hsvWLLOnW7HQ0AysOZu30x_cetMquDd1tjfyLz8HCSU,17892
|
183
|
-
ai_edge_torch/generative/layers/model_config.py,sha256=
|
183
|
+
ai_edge_torch/generative/layers/model_config.py,sha256=gGr6SZzgFVRXemfXwX__TZ0OXgAFnU70M6U0eql06TE,10712
|
184
184
|
ai_edge_torch/generative/layers/normalization.py,sha256=ijwCpi22NLX-Sygwy5sK9l9WjGvbPIhZvVwoBAonWAo,7014
|
185
185
|
ai_edge_torch/generative/layers/normalization_test.py,sha256=zwurZly-TgFxdgVVdpzu9vCpcLbd5RYt_gKg9Lfg1jI,2248
|
186
186
|
ai_edge_torch/generative/layers/rotary_position_embedding.py,sha256=975zR202MdIrILJ7blceAcxrNqX1ZCN0ECKG1gz-bV8,2655
|
@@ -215,6 +215,7 @@ ai_edge_torch/generative/utilities/model_builder.py,sha256=tBfOcsI_NcneggHqkCSyd
|
|
215
215
|
ai_edge_torch/generative/utilities/moonshine_loader.py,sha256=_RpFabSqtGH5PHiP3_1f6QfO14qMADUxr_HGRlVDFB0,4891
|
216
216
|
ai_edge_torch/generative/utilities/stable_diffusion_loader.py,sha256=dqPD9qRXEWtU3ombslOC-BE2l_dMwHoCNu7NsIJhsso,36158
|
217
217
|
ai_edge_torch/generative/utilities/t5_loader.py,sha256=tEsfy8-ymzbbjOIc-oesXF3yGyyWtJgFXn2s7VOavt8,16961
|
218
|
+
ai_edge_torch/generative/utilities/test_utils.py,sha256=fhUMCMxoeMzxYbOCjNeX5wbQmF6Y88Hi52FtRiZYJAk,1147
|
218
219
|
ai_edge_torch/generative/utilities/transformers_verifier.py,sha256=l54bmmhj613eB2oCoONIAKEHhf8TQOhC9Gwjp6lxHAE,1659
|
219
220
|
ai_edge_torch/generative/utilities/types.py,sha256=gZI9hIPB3XAo4oecKIIoVDfiyibLaSNFhecPFx4VDTM,2913
|
220
221
|
ai_edge_torch/generative/utilities/verifier.py,sha256=ETO2ShU5KXG7MLP8eVOWuzuRLCUtapafYHcZ6TZHIkw,13061
|
@@ -267,8 +268,8 @@ ai_edge_torch/testing/__init__.py,sha256=_yGgvnBZWb7T3IN3mc4x1sS4vM96HZwM8pwIcPG
|
|
267
268
|
ai_edge_torch/testing/export.py,sha256=k5mGDGzwc23Z4zaIVDs8CNh-oOt64gsf9MS9NjhbPy4,3293
|
268
269
|
ai_edge_torch/testing/model_coverage/__init__.py,sha256=5P8J6Zk5YYtDvTBucFvB9NGSRI7Gw_24WnrbhXgycEE,765
|
269
270
|
ai_edge_torch/testing/model_coverage/model_coverage.py,sha256=UPB448aMDUyC0HNYVqio2rcJPnDN0tBQMP08J6vPYew,4718
|
270
|
-
ai_edge_torch_nightly-0.6.0.
|
271
|
-
ai_edge_torch_nightly-0.6.0.
|
272
|
-
ai_edge_torch_nightly-0.6.0.
|
273
|
-
ai_edge_torch_nightly-0.6.0.
|
274
|
-
ai_edge_torch_nightly-0.6.0.
|
271
|
+
ai_edge_torch_nightly-0.6.0.dev20250530.dist-info/LICENSE,sha256=z8d0m5b2O9McPEK1xHG_dWgUBT6EfBDz6wA0F7xSPTA,11358
|
272
|
+
ai_edge_torch_nightly-0.6.0.dev20250530.dist-info/METADATA,sha256=7oeZ6wSsBUuvNXH20tOHtlWkw_Rfmmr0EADK7hSt6AQ,2074
|
273
|
+
ai_edge_torch_nightly-0.6.0.dev20250530.dist-info/WHEEL,sha256=tZoeGjtWxWRfdplE7E3d45VPlLNQnvbKiYnx7gwAy8A,92
|
274
|
+
ai_edge_torch_nightly-0.6.0.dev20250530.dist-info/top_level.txt,sha256=5KXRaF2hwkApYxf7Y8y_tVb9aulGTlbOoNdbx1aKRkE,14
|
275
|
+
ai_edge_torch_nightly-0.6.0.dev20250530.dist-info/RECORD,,
|
File without changes
|
File without changes
|