ai-edge-torch-nightly 0.3.0.dev20241002__py3-none-any.whl → 0.3.0.dev20241004__py3-none-any.whl
Sign up to get free protection for your applications and to get access to all the features.
- ai_edge_torch/generative/examples/gemma/gemma1.py +10 -93
- ai_edge_torch/generative/examples/gemma/gemma2.py +0 -1
- ai_edge_torch/generative/examples/llama/convert_to_tflite.py +13 -2
- ai_edge_torch/generative/examples/llama/llama.py +19 -24
- ai_edge_torch/generative/examples/llama/verify.py +18 -3
- ai_edge_torch/generative/examples/openelm/openelm.py +9 -90
- ai_edge_torch/generative/examples/phi/phi2.py +10 -86
- ai_edge_torch/generative/examples/phi/phi3.py +9 -69
- ai_edge_torch/generative/examples/qwen/qwen.py +26 -36
- ai_edge_torch/generative/examples/smollm/smollm.py +10 -30
- ai_edge_torch/generative/examples/tiny_llama/tiny_llama.py +11 -101
- ai_edge_torch/generative/layers/model_config.py +6 -0
- ai_edge_torch/generative/test/test_loader.py +2 -1
- ai_edge_torch/generative/test/test_model_conversion.py +39 -17
- ai_edge_torch/generative/test/test_model_conversion_large.py +6 -5
- ai_edge_torch/generative/utilities/model_builder.py +141 -0
- ai_edge_torch/version.py +1 -1
- {ai_edge_torch_nightly-0.3.0.dev20241002.dist-info → ai_edge_torch_nightly-0.3.0.dev20241004.dist-info}/METADATA +1 -1
- {ai_edge_torch_nightly-0.3.0.dev20241002.dist-info → ai_edge_torch_nightly-0.3.0.dev20241004.dist-info}/RECORD +22 -23
- ai_edge_torch/generative/examples/llama/convert_3b_to_tflite.py +0 -68
- ai_edge_torch/generative/examples/llama/verify_3b.py +0 -73
- {ai_edge_torch_nightly-0.3.0.dev20241002.dist-info → ai_edge_torch_nightly-0.3.0.dev20241004.dist-info}/LICENSE +0 -0
- {ai_edge_torch_nightly-0.3.0.dev20241002.dist-info → ai_edge_torch_nightly-0.3.0.dev20241004.dist-info}/WHEEL +0 -0
- {ai_edge_torch_nightly-0.3.0.dev20241002.dist-info → ai_edge_torch_nightly-0.3.0.dev20241004.dist-info}/top_level.txt +0 -0
@@ -3,7 +3,7 @@ ai_edge_torch/config.py,sha256=FMWeCH2b7HYILBvaI1iZNnYCO4WAhDOwBZBmIE-xrF0,909
|
|
3
3
|
ai_edge_torch/conftest.py,sha256=r0GTrhMRhlmOGrrkvumHN8hkmyug6WvF60vWq8wRIBI,758
|
4
4
|
ai_edge_torch/fx_pass_base.py,sha256=D86Gw3pIRcpnTebUPKlnPbPGJae1S6Fw4DZZ3ZkD0zw,3730
|
5
5
|
ai_edge_torch/model.py,sha256=N-pNpTxzhaFGhWhnSGd70lBzb9VlEhTOq5mddU7bvvI,5542
|
6
|
-
ai_edge_torch/version.py,sha256=
|
6
|
+
ai_edge_torch/version.py,sha256=tIC9MEJewU0lAFO_930WizESB627b7x4xfE3qbYWtLw,706
|
7
7
|
ai_edge_torch/_convert/__init__.py,sha256=hHLluseD2R0Hh4W6XZRIXY_dRQeYudjsrKGf6LZz65g,671
|
8
8
|
ai_edge_torch/_convert/conversion.py,sha256=5uPwHhmc6kwiIz-CqaiHDejf2SOWMHrb-rYEHm69wKc,3801
|
9
9
|
ai_edge_torch/_convert/conversion_utils.py,sha256=Sr8qXVcTwc-ZnZmK7yxVrIOOp1S_vNrwzC0zUvLTI2o,2160
|
@@ -41,35 +41,33 @@ ai_edge_torch/generative/examples/__init__.py,sha256=hHLluseD2R0Hh4W6XZRIXY_dRQe
|
|
41
41
|
ai_edge_torch/generative/examples/gemma/__init__.py,sha256=hHLluseD2R0Hh4W6XZRIXY_dRQeYudjsrKGf6LZz65g,671
|
42
42
|
ai_edge_torch/generative/examples/gemma/convert_gemma1_to_tflite.py,sha256=evmUj_4yygQthSRU-ke-Xn1qFNDCZKbegqINWfruKwU,2184
|
43
43
|
ai_edge_torch/generative/examples/gemma/convert_gemma2_to_tflite.py,sha256=RZDs6oY-NLYrPNtfuJDweIHzGUL2kzpIc3AW_1p8gGg,2186
|
44
|
-
ai_edge_torch/generative/examples/gemma/gemma1.py,sha256=
|
45
|
-
ai_edge_torch/generative/examples/gemma/gemma2.py,sha256=
|
44
|
+
ai_edge_torch/generative/examples/gemma/gemma1.py,sha256=oSbysiPvwp5efMbNYZop3HrxDMGiD15Tmz-HiQuTr2E,3315
|
45
|
+
ai_edge_torch/generative/examples/gemma/gemma2.py,sha256=RQFQDMEnIVp8PefcCTr7P0CvllKI7FVoIJLXbPLLIsc,9056
|
46
46
|
ai_edge_torch/generative/examples/gemma/verify_gemma1.py,sha256=ip-Gmk4CI5f0GWSdAIdrectxQWJ0t328KCsA4nfHuGg,1736
|
47
47
|
ai_edge_torch/generative/examples/gemma/verify_gemma2.py,sha256=K77k-JpdhIwm3tbBnzpw8HQsFRwAVyszxRo82fR6-q4,1762
|
48
48
|
ai_edge_torch/generative/examples/gemma/verify_util.py,sha256=sqltZbnyKemNvKqqi9d09i74gP-PPQFodRYfDfnhycQ,4933
|
49
49
|
ai_edge_torch/generative/examples/llama/__init__.py,sha256=hHLluseD2R0Hh4W6XZRIXY_dRQeYudjsrKGf6LZz65g,671
|
50
|
-
ai_edge_torch/generative/examples/llama/
|
51
|
-
ai_edge_torch/generative/examples/llama/
|
52
|
-
ai_edge_torch/generative/examples/llama/
|
53
|
-
ai_edge_torch/generative/examples/llama/verify.py,sha256=7xwKM_yzLCrmFsYj1UbsjW58ZG8Yic0xw1GFkdydrCU,2525
|
54
|
-
ai_edge_torch/generative/examples/llama/verify_3b.py,sha256=IijBWqLXINOfwayM-8EIpc7OcC6Nj5CnberStx-vDSk,2528
|
50
|
+
ai_edge_torch/generative/examples/llama/convert_to_tflite.py,sha256=P0-pByTM5tslE23ILgo7nd0nOGE25ciBRG5wKJj0bBk,2411
|
51
|
+
ai_edge_torch/generative/examples/llama/llama.py,sha256=AMcCbuDBxEfbO-l3KiEXbUaXEJ3RLLwkHii7to7UhVo,6854
|
52
|
+
ai_edge_torch/generative/examples/llama/verify.py,sha256=X7oKQi85M789ugBrOlMvzk8eSRR3Kf1Mprfl-U-WIpo,2842
|
55
53
|
ai_edge_torch/generative/examples/openelm/__init__.py,sha256=hHLluseD2R0Hh4W6XZRIXY_dRQeYudjsrKGf6LZz65g,671
|
56
54
|
ai_edge_torch/generative/examples/openelm/convert_to_tflite.py,sha256=85FVEt6cKFP2UzCLC78tAkbwGlGhAArtG7Wa75NxJik,2185
|
57
|
-
ai_edge_torch/generative/examples/openelm/openelm.py,sha256=
|
55
|
+
ai_edge_torch/generative/examples/openelm/openelm.py,sha256=JsrtuUY4q1Rovxsht2cGCuANUj1sUKnah6bAoSe8AoU,4387
|
58
56
|
ai_edge_torch/generative/examples/openelm/verify.py,sha256=VkigoqhAr8ew95neb3TifYv-SLOSheaWKv2AH0iKDrc,2441
|
59
57
|
ai_edge_torch/generative/examples/phi/__init__.py,sha256=hHLluseD2R0Hh4W6XZRIXY_dRQeYudjsrKGf6LZz65g,671
|
60
58
|
ai_edge_torch/generative/examples/phi/convert_phi3_to_tflite.py,sha256=rkbTtMaqSVG48cm-NTxR_LDgZmXAEBqayTm9O49oMXc,2171
|
61
59
|
ai_edge_torch/generative/examples/phi/convert_to_tflite.py,sha256=3go690yX6PFeXMdpY7y4JZorAwxX0HT_b_pKZieauvk,2169
|
62
|
-
ai_edge_torch/generative/examples/phi/phi2.py,sha256=
|
63
|
-
ai_edge_torch/generative/examples/phi/phi3.py,sha256=
|
60
|
+
ai_edge_torch/generative/examples/phi/phi2.py,sha256=CQ55KfOdoOM43CxF7yNQsgq8b-j0S50bXpxYzgq-keM,3418
|
61
|
+
ai_edge_torch/generative/examples/phi/phi3.py,sha256=GkHOaYfsFEbHvfZCaLlb3Us_h19ezqPDUakoz_DiG9A,7123
|
64
62
|
ai_edge_torch/generative/examples/phi/verify.py,sha256=YPFCdbnfmvq38fbpBNr0kHPfSZo4p3_6WkLJAW3pLPo,2177
|
65
63
|
ai_edge_torch/generative/examples/phi/verify_phi3.py,sha256=kVYaBVvddfQng0IyZGxyTJEzhiPO0G4VFJm2WOc2Q94,2360
|
66
64
|
ai_edge_torch/generative/examples/qwen/__init__.py,sha256=hHLluseD2R0Hh4W6XZRIXY_dRQeYudjsrKGf6LZz65g,671
|
67
65
|
ai_edge_torch/generative/examples/qwen/convert_to_tflite.py,sha256=QAAVoSKDVf2rHAChzumGloVCWIU0Oe5UYKgv3T192Iw,2496
|
68
|
-
ai_edge_torch/generative/examples/qwen/qwen.py,sha256=
|
66
|
+
ai_edge_torch/generative/examples/qwen/qwen.py,sha256=oYm9hhALUQ4uOn-PO1bF7fCIGP8EWRNK4zClkx2RQs8,4070
|
69
67
|
ai_edge_torch/generative/examples/qwen/verify.py,sha256=9_AyEJTeUfvhhID64Rto2bflFPyXMFokdQLsseLUMiI,2775
|
70
68
|
ai_edge_torch/generative/examples/smollm/__init__.py,sha256=hHLluseD2R0Hh4W6XZRIXY_dRQeYudjsrKGf6LZz65g,671
|
71
69
|
ai_edge_torch/generative/examples/smollm/convert_to_tflite.py,sha256=zPrDTDeRVWFi9DS32uNi-RLpzOStFOk5MhNla4ixeew,2179
|
72
|
-
ai_edge_torch/generative/examples/smollm/smollm.py,sha256=
|
70
|
+
ai_edge_torch/generative/examples/smollm/smollm.py,sha256=M5qAcSUE5gxOSfq24a8lZku9kgvmlFCyIBar3kF2XEk,2570
|
73
71
|
ai_edge_torch/generative/examples/smollm/verify.py,sha256=HXYcCjDJMylVL3Pc9HU-UXqtpjtIU25o1YhPiX30aPU,2361
|
74
72
|
ai_edge_torch/generative/examples/stable_diffusion/__init__.py,sha256=hHLluseD2R0Hh4W6XZRIXY_dRQeYudjsrKGf6LZz65g,671
|
75
73
|
ai_edge_torch/generative/examples/stable_diffusion/attention.py,sha256=kDWG6MlIGa89zC5KSRcJlw2c4ITuw8KcchtfmF55f4g,3545
|
@@ -96,7 +94,7 @@ ai_edge_torch/generative/examples/test_models/toy_model.py,sha256=4113jZK-Hu3kYo
|
|
96
94
|
ai_edge_torch/generative/examples/test_models/toy_model_with_kv_cache.py,sha256=ZpjSIiayjTEVwg5Q1vI9Iy5tq1YSF5zaVDF4HTp_Z2s,4353
|
97
95
|
ai_edge_torch/generative/examples/tiny_llama/__init__.py,sha256=hHLluseD2R0Hh4W6XZRIXY_dRQeYudjsrKGf6LZz65g,671
|
98
96
|
ai_edge_torch/generative/examples/tiny_llama/convert_to_tflite.py,sha256=ekxd8efjMgEvauUu3PidWOC-DszPHn5sqU753F7sJIM,2201
|
99
|
-
ai_edge_torch/generative/examples/tiny_llama/tiny_llama.py,sha256=
|
97
|
+
ai_edge_torch/generative/examples/tiny_llama/tiny_llama.py,sha256=10X8HwPx4akzclnIMOBNItKQemhRbvxBbTo7nwZtWjM,2650
|
100
98
|
ai_edge_torch/generative/examples/tiny_llama/verify.py,sha256=7Bk8z033M-BCXJ299fpQNXYAudBbZoDQp9934xcvg50,2426
|
101
99
|
ai_edge_torch/generative/fx_passes/__init__.py,sha256=jrzCB3ZyY_t5jJM1e2Czdt3DjAIL43R0_a-T-I7wOzw,1155
|
102
100
|
ai_edge_torch/generative/fx_passes/remove_sdpa_zero_mask_pass.py,sha256=hhxSQvkDMv0isZJhmuLiod66ZODaJ8uSPSVTJVHBabQ,1931
|
@@ -106,7 +104,7 @@ ai_edge_torch/generative/layers/attention_utils.py,sha256=68GXGR2HSWBFViTxX7cHif
|
|
106
104
|
ai_edge_torch/generative/layers/builder.py,sha256=oE8DdqLA-oWkBC2zySSCh8JNAJg_hk8-W_UoMSrgDVk,5088
|
107
105
|
ai_edge_torch/generative/layers/feed_forward.py,sha256=hdICat-8gW7-vxDAevJQ8NQ-mynllPiqLdXQMF6JMnc,4189
|
108
106
|
ai_edge_torch/generative/layers/kv_cache.py,sha256=2El7kZYnQRCRcVc63xgiAdBh9oVOksDu35p9XggvaGE,6148
|
109
|
-
ai_edge_torch/generative/layers/model_config.py,sha256=
|
107
|
+
ai_edge_torch/generative/layers/model_config.py,sha256=xZt4xaNZJPvtdy4hfbnRencEENr689zO0WnZbhpNTIs,7137
|
110
108
|
ai_edge_torch/generative/layers/normalization.py,sha256=cpo88JUXbF9j3sJTU4JuwOap9ryGV05C1QkPij-YQwU,6999
|
111
109
|
ai_edge_torch/generative/layers/rotary_position_embedding.py,sha256=CZqOoibLcHvUgrgaIIWAlmk3XgE2inzx340MN-npLoU,1347
|
112
110
|
ai_edge_torch/generative/layers/scaled_dot_product_attention.py,sha256=gXxh3papKy4FBpGEX7VyZ7rZ1Js6aHK70Q6DKrVSckY,4154
|
@@ -123,14 +121,15 @@ ai_edge_torch/generative/quantize/quant_recipes.py,sha256=0Kvr_o7pbMnE8VMe6Ml0FB
|
|
123
121
|
ai_edge_torch/generative/quantize/supported_schemes.py,sha256=FjdycEOvxRgBmQdZVufetPvkDoD7rUowIOSKV9oV5Kk,1418
|
124
122
|
ai_edge_torch/generative/test/__init__.py,sha256=hHLluseD2R0Hh4W6XZRIXY_dRQeYudjsrKGf6LZz65g,671
|
125
123
|
ai_edge_torch/generative/test/test_kv_cache.py,sha256=W6Bh0gYDzmwb0j9HdD5_D7Z7FPToP2HSyFrmwIXuFqo,3793
|
126
|
-
ai_edge_torch/generative/test/test_loader.py,sha256=
|
127
|
-
ai_edge_torch/generative/test/test_model_conversion.py,sha256=
|
128
|
-
ai_edge_torch/generative/test/test_model_conversion_large.py,sha256=
|
124
|
+
ai_edge_torch/generative/test/test_loader.py,sha256=9mQUeeZKOVApOWSWl2cN9c10axZjMKM1-0Zd823CCS4,3449
|
125
|
+
ai_edge_torch/generative/test/test_model_conversion.py,sha256=a4TzSw8KMxEafirxqkykZi-WgTs5Z7wHp-J1AfjRDzA,6353
|
126
|
+
ai_edge_torch/generative/test/test_model_conversion_large.py,sha256=bVCm_mubuGszCBON6oRjQXcBgPZqlVmmOaLWwhZJLio,9060
|
129
127
|
ai_edge_torch/generative/test/test_quantize.py,sha256=8geJhKwYBU20m0mdGPD1BUFwQ0lZKNtCB04SOLO18y4,5980
|
130
128
|
ai_edge_torch/generative/test/utils.py,sha256=YvEhO2HIj1LkBs5du1UxY-cGRW9HMyAYsOUhgsTrTpA,1796
|
131
129
|
ai_edge_torch/generative/utilities/__init__.py,sha256=-_jxnnFnCgnTU4oTm4MnRsvL5lqhomBNdFBbqfmfHPo,720
|
132
130
|
ai_edge_torch/generative/utilities/converter.py,sha256=MQUg2ZLmfk_2csWmQWKD_II0bXq4X3McI5i-qWraieE,2987
|
133
131
|
ai_edge_torch/generative/utilities/loader.py,sha256=b9iotIhVDX-Zc9XjIDUaLxnV395AyBnkQe3dV5YA7Co,13297
|
132
|
+
ai_edge_torch/generative/utilities/model_builder.py,sha256=89jt80UUfDzYBi-x077HBavWeuNJuYPXym9fiKCY1Tk,5278
|
134
133
|
ai_edge_torch/generative/utilities/stable_diffusion_loader.py,sha256=dqPD9qRXEWtU3ombslOC-BE2l_dMwHoCNu7NsIJhsso,36158
|
135
134
|
ai_edge_torch/generative/utilities/t5_loader.py,sha256=tEsfy8-ymzbbjOIc-oesXF3yGyyWtJgFXn2s7VOavt8,16961
|
136
135
|
ai_edge_torch/generative/utilities/transformers_verifier.py,sha256=8sp9m_FMcXn7nqOrochtu2jIANkJKhnhIBUmH0ZTDR4,1549
|
@@ -181,8 +180,8 @@ ai_edge_torch/quantize/quant_config.py,sha256=U0KisSW-uZkoMJcy-ZP9W57p3tsa594fr9
|
|
181
180
|
ai_edge_torch/testing/__init__.py,sha256=hHLluseD2R0Hh4W6XZRIXY_dRQeYudjsrKGf6LZz65g,671
|
182
181
|
ai_edge_torch/testing/model_coverage/__init__.py,sha256=5P8J6Zk5YYtDvTBucFvB9NGSRI7Gw_24WnrbhXgycEE,765
|
183
182
|
ai_edge_torch/testing/model_coverage/model_coverage.py,sha256=UPB448aMDUyC0HNYVqio2rcJPnDN0tBQMP08J6vPYew,4718
|
184
|
-
ai_edge_torch_nightly-0.3.0.
|
185
|
-
ai_edge_torch_nightly-0.3.0.
|
186
|
-
ai_edge_torch_nightly-0.3.0.
|
187
|
-
ai_edge_torch_nightly-0.3.0.
|
188
|
-
ai_edge_torch_nightly-0.3.0.
|
183
|
+
ai_edge_torch_nightly-0.3.0.dev20241004.dist-info/LICENSE,sha256=z8d0m5b2O9McPEK1xHG_dWgUBT6EfBDz6wA0F7xSPTA,11358
|
184
|
+
ai_edge_torch_nightly-0.3.0.dev20241004.dist-info/METADATA,sha256=LZEnnjuiIFRFASjn-R5mEPu8juBMx7ZvLgbGZuv9CQw,1897
|
185
|
+
ai_edge_torch_nightly-0.3.0.dev20241004.dist-info/WHEEL,sha256=eOLhNAGa2EW3wWl_TU484h7q1UNgy0JXjjoqKoxAAQc,92
|
186
|
+
ai_edge_torch_nightly-0.3.0.dev20241004.dist-info/top_level.txt,sha256=5KXRaF2hwkApYxf7Y8y_tVb9aulGTlbOoNdbx1aKRkE,14
|
187
|
+
ai_edge_torch_nightly-0.3.0.dev20241004.dist-info/RECORD,,
|
@@ -1,68 +0,0 @@
|
|
1
|
-
# Copyright 2024 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
|
-
|
16
|
-
"""Example of converting Llama 3.2 3B model to multi-signature tflite model."""
|
17
|
-
|
18
|
-
import os
|
19
|
-
import pathlib
|
20
|
-
|
21
|
-
from absl import app
|
22
|
-
from absl import flags
|
23
|
-
from ai_edge_torch.generative.examples.llama import llama
|
24
|
-
from ai_edge_torch.generative.utilities import converter
|
25
|
-
|
26
|
-
_CHECKPOINT_PATH = flags.DEFINE_string(
|
27
|
-
'checkpoint_path',
|
28
|
-
os.path.join(pathlib.Path.home(), 'Downloads/llm_data/llama'),
|
29
|
-
'The path to the model checkpoint, or directory holding the checkpoint.',
|
30
|
-
)
|
31
|
-
_TFLITE_PATH = flags.DEFINE_string(
|
32
|
-
'tflite_path',
|
33
|
-
'/tmp/',
|
34
|
-
'The tflite file path to export.',
|
35
|
-
)
|
36
|
-
_PREFILL_SEQ_LEN = flags.DEFINE_integer(
|
37
|
-
'prefill_seq_len',
|
38
|
-
1024,
|
39
|
-
'The maximum size of prefill input tensor.',
|
40
|
-
)
|
41
|
-
_KV_CACHE_MAX_LEN = flags.DEFINE_integer(
|
42
|
-
'kv_cache_max_len',
|
43
|
-
1280,
|
44
|
-
'The maximum size of KV cache buffer, including both prefill and decode.',
|
45
|
-
)
|
46
|
-
_QUANTIZE = flags.DEFINE_bool(
|
47
|
-
'quantize',
|
48
|
-
True,
|
49
|
-
'Whether the model should be quantized.',
|
50
|
-
)
|
51
|
-
|
52
|
-
|
53
|
-
def main(_):
|
54
|
-
pytorch_model = llama.build_3b_model(
|
55
|
-
_CHECKPOINT_PATH.value, kv_cache_max_len=_KV_CACHE_MAX_LEN.value
|
56
|
-
)
|
57
|
-
quant_suffix = 'q8' if _QUANTIZE.value else 'f32'
|
58
|
-
output_filename = f'llama_3b_{quant_suffix}_seq{_PREFILL_SEQ_LEN.value}_ekv{_KV_CACHE_MAX_LEN.value}.tflite'
|
59
|
-
converter.convert_to_tflite(
|
60
|
-
pytorch_model,
|
61
|
-
tflite_path=os.path.join(_TFLITE_PATH.value, output_filename),
|
62
|
-
prefill_seq_len=_PREFILL_SEQ_LEN.value,
|
63
|
-
quantize=_QUANTIZE.value,
|
64
|
-
)
|
65
|
-
|
66
|
-
|
67
|
-
if __name__ == '__main__':
|
68
|
-
app.run(main)
|
@@ -1,73 +0,0 @@
|
|
1
|
-
# Copyright 2024 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
|
-
|
16
|
-
"""Verifies the reauthored Llama 3.2-3B model."""
|
17
|
-
|
18
|
-
import logging
|
19
|
-
import pathlib
|
20
|
-
|
21
|
-
from absl import app
|
22
|
-
from absl import flags
|
23
|
-
from ai_edge_torch.generative.examples.llama import llama
|
24
|
-
from ai_edge_torch.generative.utilities import transformers_verifier
|
25
|
-
from ai_edge_torch.generative.utilities import verifier
|
26
|
-
import transformers
|
27
|
-
|
28
|
-
|
29
|
-
_PROMPTS = flags.DEFINE_multi_string(
|
30
|
-
"prompts",
|
31
|
-
"What is the meaning of life?",
|
32
|
-
"The input prompts to generate answers.",
|
33
|
-
)
|
34
|
-
_MAX_NEW_TOKENS = flags.DEFINE_integer(
|
35
|
-
"max_new_tokens",
|
36
|
-
30,
|
37
|
-
"The maximum size of the generated tokens.",
|
38
|
-
)
|
39
|
-
|
40
|
-
|
41
|
-
def main(_):
|
42
|
-
checkpoint = "meta-llama/Llama-3.2-3B-Instruct"
|
43
|
-
logging.info("Loading the original model from: %s", checkpoint)
|
44
|
-
original_model = transformers.AutoModelForCausalLM.from_pretrained(checkpoint)
|
45
|
-
|
46
|
-
# Locate the cached dir.
|
47
|
-
cached_config_file = transformers.utils.cached_file(
|
48
|
-
checkpoint, transformers.utils.CONFIG_NAME
|
49
|
-
)
|
50
|
-
reauthored_checkpoint = pathlib.Path(cached_config_file).parent
|
51
|
-
logging.info("Building the reauthored model from: %s", reauthored_checkpoint)
|
52
|
-
reauthored_model = llama.build_3b_model(reauthored_checkpoint)
|
53
|
-
|
54
|
-
logging.info("Loading the tokenizer from: %s", checkpoint)
|
55
|
-
# Llama tokenizer_config.json sets a fast tokenizer class explicitly,
|
56
|
-
# "PreTrainedTokenizerFast". It works only when the fast tokenizer is
|
57
|
-
# available.
|
58
|
-
tokenizer = transformers.AutoTokenizer.from_pretrained(checkpoint)
|
59
|
-
|
60
|
-
verifier.verify_reauthored_model(
|
61
|
-
original_model=transformers_verifier.TransformersModelWrapper(
|
62
|
-
original_model
|
63
|
-
),
|
64
|
-
reauthored_model=verifier.ReauthoredModelWrapper(reauthored_model),
|
65
|
-
tokenizer=verifier.TokenizerWrapper(tokenizer),
|
66
|
-
generate_prompts=_PROMPTS.value,
|
67
|
-
max_new_tokens=_MAX_NEW_TOKENS.value,
|
68
|
-
atol=1e-04,
|
69
|
-
)
|
70
|
-
|
71
|
-
|
72
|
-
if __name__ == "__main__":
|
73
|
-
app.run(main)
|
File without changes
|
File without changes
|