onnxruntime 0.7.7-x86_64-darwin → 0.8.0-x86_64-darwin

Sign up to get free protection for your applications and to get access to all the features.
checksums.yaml CHANGED
@@ -1,7 +1,7 @@
1
1
  ---
2
2
  SHA256:
3
- metadata.gz: 6bf68a601b76d9895abd8d0ae00c73b084ca1c9cd316e38c43d4a3ded7d83965
4
- data.tar.gz: 6b81ff6a468fb51cd4385098f3d17e291f75259d6b293d0c6c174eb6c29427e7
3
+ metadata.gz: b7bb1d0a96ee5414fe50ef76fba8db958a679f81bb1f378523eba9660f12c466
4
+ data.tar.gz: 44c56ae37a5178cd6db8a4bb6f16114a27fab6202f60bf7cba280ef653525a11
5
5
  SHA512:
6
- metadata.gz: 9bc8b3a8b720c57f93c6f0e9fd793b529c7445bb4dd82501697fe55ea9888f800400042671e70b101b7f94f0bcac132e52ab43baf9c274d5c99b7ad8d72db95a
7
- data.tar.gz: 3b33faa125e0d9613891c0fdaed7b07ba53636cb63268f57fac9f667687e7219db4ff9cc0b36573689a779876c2b50d0c5075a0a1ee847856ecca46e7b21bc94
6
+ metadata.gz: bc4fcd2da365002b9c833fa0062b360dc0b2dc48a6fbcde6286539a75fabe375f544793ce83a4f3b93354c685ff56a529b5f9bc93c1b1c7249437f4110b3c34c
7
+ data.tar.gz: be4534d7cb1ba210cc818004b017583de8e4c4d48ccccd5fcdfabd6962057d6d5ebe05b1c9067a96b059fbf6a2d97dea3dd2e48057f912113df3c730a186d7f4
data/CHANGELOG.md CHANGED
@@ -1,3 +1,11 @@
1
+ ## 0.8.0 (2023-09-20)
2
+
3
+ - Updated ONNX Runtime to 1.16.0
4
+ - Changed inputs and outputs to return symbolic dimension names
5
+ - Fixed GPU support
6
+ - Dropped support for Ruby < 3
7
+ - Dropped support for loading models from binary string (use `StringIO` instead)
8
+
1
9
  ## 0.7.7 (2023-07-24)
2
10
 
3
11
  - Updated ONNX Runtime to 1.15.1
data/README.md CHANGED
@@ -114,6 +114,12 @@ To enable GPU support on Linux and Windows, download the appropriate [GPU releas
114
114
  OnnxRuntime.ffi_lib = "path/to/lib/libonnxruntime.so" # onnxruntime.dll for Windows
115
115
  ```
116
116
 
117
+ and use:
118
+
119
+ ```ruby
120
+ model = OnnxRuntime::Model.new("model.onnx", providers: ["CUDAExecutionProvider"])
121
+ ```
122
+
117
123
  ## History
118
124
 
119
125
  View the [changelog](https://github.com/ankane/onnxruntime-ruby/blob/master/CHANGELOG.md)
@@ -7,7 +7,7 @@ module OnnxRuntime
7
7
  # https://github.com/microsoft/onnxruntime/blob/master/include/onnxruntime/core/session/onnxruntime_c_api.h
8
8
  # keep same order
9
9
 
10
- ORT_API_VERSION = 11
10
+ ORT_API_VERSION = 16
11
11
 
12
12
  # enums
13
13
  TensorElementDataType = enum(:undefined, :float, :uint8, :int8, :uint16, :int16, :int32, :int64, :string, :bool, :float16, :double, :uint32, :uint64, :complex64, :complex128, :bfloat16)
@@ -199,6 +199,7 @@ module OnnxRuntime
199
199
  :FillSparseTensorCsr, callback(%i[], :pointer),
200
200
  :FillSparseTensorBlockSparse, callback(%i[], :pointer),
201
201
  :CreateSparseTensorWithValuesAsOrtValue, callback(%i[], :pointer),
202
+ :UseCooIndices, callback(%i[], :pointer),
202
203
  :UseCsrIndices, callback(%i[], :pointer),
203
204
  :UseBlockSparseIndices, callback(%i[], :pointer),
204
205
  :GetSparseTensorFormat, callback(%i[], :pointer),
@@ -218,11 +219,11 @@ module OnnxRuntime
218
219
  :SetGlobalCustomJoinThreadFn, callback(%i[], :pointer),
219
220
  :SynchronizeBoundInputs, callback(%i[], :pointer),
220
221
  :SynchronizeBoundOutputs, callback(%i[], :pointer),
221
- :SessionOptionsAppendExecutionProvider_CUDA_V2, callback(%i[], :pointer),
222
- :CreateCUDAProviderOptions, callback(%i[], :pointer),
223
- :UpdateCUDAProviderOptions, callback(%i[], :pointer),
224
- :GetCUDAProviderOptionsAsString, callback(%i[], :pointer),
225
- :ReleaseCUDAProviderOptions, callback(%i[], :pointer),
222
+ :SessionOptionsAppendExecutionProvider_CUDA_V2, callback(%i[pointer pointer], :pointer),
223
+ :CreateCUDAProviderOptions, callback(%i[pointer], :pointer),
224
+ :UpdateCUDAProviderOptions, callback(%i[pointer pointer pointer size_t], :pointer),
225
+ :GetCUDAProviderOptionsAsString, callback(%i[pointer pointer pointer], :pointer),
226
+ :ReleaseCUDAProviderOptions, callback(%i[pointer], :void),
226
227
  :SessionOptionsAppendExecutionProvider_MIGraphX, callback(%i[], :pointer)
227
228
  end
228
229
 
@@ -2,7 +2,7 @@ module OnnxRuntime
2
2
  class InferenceSession
3
3
  attr_reader :inputs, :outputs
4
4
 
5
- def initialize(path_or_bytes, enable_cpu_mem_arena: true, enable_mem_pattern: true, enable_profiling: false, execution_mode: nil, free_dimension_overrides_by_denotation: nil, free_dimension_overrides_by_name: nil, graph_optimization_level: nil, inter_op_num_threads: nil, intra_op_num_threads: nil, log_severity_level: nil, log_verbosity_level: nil, logid: nil, optimized_model_filepath: nil, profile_file_prefix: nil, session_config_entries: nil)
5
+ def initialize(path_or_bytes, enable_cpu_mem_arena: true, enable_mem_pattern: true, enable_profiling: false, execution_mode: nil, free_dimension_overrides_by_denotation: nil, free_dimension_overrides_by_name: nil, graph_optimization_level: nil, inter_op_num_threads: nil, intra_op_num_threads: nil, log_severity_level: nil, log_verbosity_level: nil, logid: nil, optimized_model_filepath: nil, profile_file_prefix: nil, session_config_entries: nil, providers: [])
6
6
  # session options
7
7
  session_options = ::FFI::MemoryPointer.new(:pointer)
8
8
  check_status api[:CreateSessionOptions].call(session_options)
@@ -54,6 +54,24 @@ module OnnxRuntime
54
54
  check_status api[:AddSessionConfigEntry].call(session_options.read_pointer, k.to_s, v.to_s)
55
55
  end
56
56
  end
57
+ providers.each do |provider|
58
+ unless self.providers.include?(provider)
59
+ warn "Provider not available: #{provider}"
60
+ next
61
+ end
62
+
63
+ case provider
64
+ when "CUDAExecutionProvider"
65
+ cuda_options = ::FFI::MemoryPointer.new(:pointer)
66
+ check_status api[:CreateCUDAProviderOptions].call(cuda_options)
67
+ check_status api[:SessionOptionsAppendExecutionProvider_CUDA_V2].call(session_options.read_pointer, cuda_options.read_pointer)
68
+ release :CUDAProviderOptions, cuda_options
69
+ when "CPUExecutionProvider"
70
+ break
71
+ else
72
+ raise ArgumentError, "Provider not supported: #{provider}"
73
+ end
74
+ end
57
75
 
58
76
  @session = load_session(path_or_bytes, session_options)
59
77
  ObjectSpace.define_finalizer(@session, self.class.finalize(read_pointer.to_i))
@@ -185,8 +203,7 @@ module OnnxRuntime
185
203
  true
186
204
  else
187
205
  path_or_bytes = path_or_bytes.to_str
188
- # TODO remove ability to load byte string directly in 0.8.0
189
- path_or_bytes.encoding == Encoding::BINARY
206
+ false
190
207
  end
191
208
 
192
209
  if from_memory
@@ -509,11 +526,10 @@ module OnnxRuntime
509
526
  check_status api[:GetDimensions].call(tensor_info.read_pointer, node_dims, num_dims)
510
527
  dims = node_dims.read_array_of_int64(num_dims)
511
528
 
512
- # TODO uncomment in 0.8.0
513
- # symbolic_dims = ::FFI::MemoryPointer.new(:pointer, num_dims)
514
- # check_status api[:GetSymbolicDimensions].call(tensor_info.read_pointer, symbolic_dims, num_dims)
515
- # named_dims = num_dims.times.map { |i| symbolic_dims[i].read_pointer.read_string }
516
- # dims = named_dims.zip(dims).map { |n, d| n.empty? ? d : n }
529
+ symbolic_dims = ::FFI::MemoryPointer.new(:pointer, num_dims)
530
+ check_status api[:GetSymbolicDimensions].call(tensor_info.read_pointer, symbolic_dims, num_dims)
531
+ named_dims = num_dims.times.map { |i| symbolic_dims[i].read_pointer.read_string }
532
+ dims = named_dims.zip(dims).map { |n, d| n.empty? ? d : n }
517
533
 
518
534
  [type.read_int, dims]
519
535
  end
@@ -1,3 +1,3 @@
1
1
  module OnnxRuntime
2
- VERSION = "0.7.7"
2
+ VERSION = "0.8.0"
3
3
  end
@@ -6021,4 +6021,246 @@ OR OTHER DEALINGS IN THE SOFTWARE.
6021
6021
 
6022
6022
  Except as contained in this notice, the name of a copyright holder shall not
6023
6023
  be used in advertising or otherwise to promote the sale, use or other dealings
6024
- in this Software without prior written authorization of the copyright holder.
6024
+ in this Software without prior written authorization of the copyright holder.
6025
+
6026
+ _____
6027
+
6028
+ Intel neural-compressor
6029
+
6030
+ https://github.com/intel/neural-compressor
6031
+
6032
+ Apache License
6033
+ Version 2.0, January 2004
6034
+ http://www.apache.org/licenses/
6035
+
6036
+ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
6037
+
6038
+ 1. Definitions.
6039
+
6040
+ "License" shall mean the terms and conditions for use, reproduction,
6041
+ and distribution as defined by Sections 1 through 9 of this document.
6042
+
6043
+ "Licensor" shall mean the copyright owner or entity authorized by
6044
+ the copyright owner that is granting the License.
6045
+
6046
+ "Legal Entity" shall mean the union of the acting entity and all
6047
+ other entities that control, are controlled by, or are under common
6048
+ control with that entity. For the purposes of this definition,
6049
+ "control" means (i) the power, direct or indirect, to cause the
6050
+ direction or management of such entity, whether by contract or
6051
+ otherwise, or (ii) ownership of fifty percent (50%) or more of the
6052
+ outstanding shares, or (iii) beneficial ownership of such entity.
6053
+
6054
+ "You" (or "Your") shall mean an individual or Legal Entity
6055
+ exercising permissions granted by this License.
6056
+
6057
+ "Source" form shall mean the preferred form for making modifications,
6058
+ including but not limited to software source code, documentation
6059
+ source, and configuration files.
6060
+
6061
+ "Object" form shall mean any form resulting from mechanical
6062
+ transformation or translation of a Source form, including but
6063
+ not limited to compiled object code, generated documentation,
6064
+ and conversions to other media types.
6065
+
6066
+ "Work" shall mean the work of authorship, whether in Source or
6067
+ Object form, made available under the License, as indicated by a
6068
+ copyright notice that is included in or attached to the work
6069
+ (an example is provided in the Appendix below).
6070
+
6071
+ "Derivative Works" shall mean any work, whether in Source or Object
6072
+ form, that is based on (or derived from) the Work and for which the
6073
+ editorial revisions, annotations, elaborations, or other modifications
6074
+ represent, as a whole, an original work of authorship. For the purposes
6075
+ of this License, Derivative Works shall not include works that remain
6076
+ separable from, or merely link (or bind by name) to the interfaces of,
6077
+ the Work and Derivative Works thereof.
6078
+
6079
+ "Contribution" shall mean any work of authorship, including
6080
+ the original version of the Work and any modifications or additions
6081
+ to that Work or Derivative Works thereof, that is intentionally
6082
+ submitted to Licensor for inclusion in the Work by the copyright owner
6083
+ or by an individual or Legal Entity authorized to submit on behalf of
6084
+ the copyright owner. For the purposes of this definition, "submitted"
6085
+ means any form of electronic, verbal, or written communication sent
6086
+ to the Licensor or its representatives, including but not limited to
6087
+ communication on electronic mailing lists, source code control systems,
6088
+ and issue tracking systems that are managed by, or on behalf of, the
6089
+ Licensor for the purpose of discussing and improving the Work, but
6090
+ excluding communication that is conspicuously marked or otherwise
6091
+ designated in writing by the copyright owner as "Not a Contribution."
6092
+
6093
+ "Contributor" shall mean Licensor and any individual or Legal Entity
6094
+ on behalf of whom a Contribution has been received by Licensor and
6095
+ subsequently incorporated within the Work.
6096
+
6097
+ 2. Grant of Copyright License. Subject to the terms and conditions of
6098
+ this License, each Contributor hereby grants to You a perpetual,
6099
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
6100
+ copyright license to reproduce, prepare Derivative Works of,
6101
+ publicly display, publicly perform, sublicense, and distribute the
6102
+ Work and such Derivative Works in Source or Object form.
6103
+
6104
+ 3. Grant of Patent License. Subject to the terms and conditions of
6105
+ this License, each Contributor hereby grants to You a perpetual,
6106
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
6107
+ (except as stated in this section) patent license to make, have made,
6108
+ use, offer to sell, sell, import, and otherwise transfer the Work,
6109
+ where such license applies only to those patent claims licensable
6110
+ by such Contributor that are necessarily infringed by their
6111
+ Contribution(s) alone or by combination of their Contribution(s)
6112
+ with the Work to which such Contribution(s) was submitted. If You
6113
+ institute patent litigation against any entity (including a
6114
+ cross-claim or counterclaim in a lawsuit) alleging that the Work
6115
+ or a Contribution incorporated within the Work constitutes direct
6116
+ or contributory patent infringement, then any patent licenses
6117
+ granted to You under this License for that Work shall terminate
6118
+ as of the date such litigation is filed.
6119
+
6120
+ 4. Redistribution. You may reproduce and distribute copies of the
6121
+ Work or Derivative Works thereof in any medium, with or without
6122
+ modifications, and in Source or Object form, provided that You
6123
+ meet the following conditions:
6124
+
6125
+ (a) You must give any other recipients of the Work or
6126
+ Derivative Works a copy of this License; and
6127
+
6128
+ (b) You must cause any modified files to carry prominent notices
6129
+ stating that You changed the files; and
6130
+
6131
+ (c) You must retain, in the Source form of any Derivative Works
6132
+ that You distribute, all copyright, patent, trademark, and
6133
+ attribution notices from the Source form of the Work,
6134
+ excluding those notices that do not pertain to any part of
6135
+ the Derivative Works; and
6136
+
6137
+ (d) If the Work includes a "NOTICE" text file as part of its
6138
+ distribution, then any Derivative Works that You distribute must
6139
+ include a readable copy of the attribution notices contained
6140
+ within such NOTICE file, excluding those notices that do not
6141
+ pertain to any part of the Derivative Works, in at least one
6142
+ of the following places: within a NOTICE text file distributed
6143
+ as part of the Derivative Works; within the Source form or
6144
+ documentation, if provided along with the Derivative Works; or,
6145
+ within a display generated by the Derivative Works, if and
6146
+ wherever such third-party notices normally appear. The contents
6147
+ of the NOTICE file are for informational purposes only and
6148
+ do not modify the License. You may add Your own attribution
6149
+ notices within Derivative Works that You distribute, alongside
6150
+ or as an addendum to the NOTICE text from the Work, provided
6151
+ that such additional attribution notices cannot be construed
6152
+ as modifying the License.
6153
+
6154
+ You may add Your own copyright statement to Your modifications and
6155
+ may provide additional or different license terms and conditions
6156
+ for use, reproduction, or distribution of Your modifications, or
6157
+ for any such Derivative Works as a whole, provided Your use,
6158
+ reproduction, and distribution of the Work otherwise complies with
6159
+ the conditions stated in this License.
6160
+
6161
+ 5. Submission of Contributions. Unless You explicitly state otherwise,
6162
+ any Contribution intentionally submitted for inclusion in the Work
6163
+ by You to the Licensor shall be under the terms and conditions of
6164
+ this License, without any additional terms or conditions.
6165
+ Notwithstanding the above, nothing herein shall supersede or modify
6166
+ the terms of any separate license agreement you may have executed
6167
+ with Licensor regarding such Contributions.
6168
+
6169
+ 6. Trademarks. This License does not grant permission to use the trade
6170
+ names, trademarks, service marks, or product names of the Licensor,
6171
+ except as required for reasonable and customary use in describing the
6172
+ origin of the Work and reproducing the content of the NOTICE file.
6173
+
6174
+ 7. Disclaimer of Warranty. Unless required by applicable law or
6175
+ agreed to in writing, Licensor provides the Work (and each
6176
+ Contributor provides its Contributions) on an "AS IS" BASIS,
6177
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
6178
+ implied, including, without limitation, any warranties or conditions
6179
+ of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
6180
+ PARTICULAR PURPOSE. You are solely responsible for determining the
6181
+ appropriateness of using or redistributing the Work and assume any
6182
+ risks associated with Your exercise of permissions under this License.
6183
+
6184
+ 8. Limitation of Liability. In no event and under no legal theory,
6185
+ whether in tort (including negligence), contract, or otherwise,
6186
+ unless required by applicable law (such as deliberate and grossly
6187
+ negligent acts) or agreed to in writing, shall any Contributor be
6188
+ liable to You for damages, including any direct, indirect, special,
6189
+ incidental, or consequential damages of any character arising as a
6190
+ result of this License or out of the use or inability to use the
6191
+ Work (including but not limited to damages for loss of goodwill,
6192
+ work stoppage, computer failure or malfunction, or any and all
6193
+ other commercial damages or losses), even if such Contributor
6194
+ has been advised of the possibility of such damages.
6195
+
6196
+ 9. Accepting Warranty or Additional Liability. While redistributing
6197
+ the Work or Derivative Works thereof, You may choose to offer,
6198
+ and charge a fee for, acceptance of support, warranty, indemnity,
6199
+ or other liability obligations and/or rights consistent with this
6200
+ License. However, in accepting such obligations, You may act only
6201
+ on Your own behalf and on Your sole responsibility, not on behalf
6202
+ of any other Contributor, and only if You agree to indemnify,
6203
+ defend, and hold each Contributor harmless for any liability
6204
+ incurred by, or claims asserted against, such Contributor by reason
6205
+ of your accepting any such warranty or additional liability.
6206
+
6207
+ END OF TERMS AND CONDITIONS
6208
+
6209
+ ============================================================================
6210
+
6211
+ Copyright 2016-2019 Intel Corporation
6212
+ Copyright 2018 YANDEX LLC
6213
+
6214
+ Licensed under the Apache License, Version 2.0 (the "License");
6215
+ you may not use this file except in compliance with the License.
6216
+ You may obtain a copy of the License at
6217
+
6218
+ http://www.apache.org/licenses/LICENSE-2.0
6219
+
6220
+ Unless required by applicable law or agreed to in writing, software
6221
+ distributed under the License is distributed on an "AS IS" BASIS,
6222
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
6223
+ See the License for the specific language governing permissions and
6224
+ limitations under the License.
6225
+
6226
+ This distribution includes third party software ("third party programs").
6227
+ This third party software, even if included with the distribution of
6228
+ the Intel software, may be governed by separate license terms, including
6229
+ without limitation, third party license terms, other Intel software license
6230
+ terms, and open source software license terms. These separate license terms
6231
+ govern your use of the third party programs as set forth in the
6232
+ "THIRD-PARTY-PROGRAMS" file.
6233
+
6234
+ _____
6235
+
6236
+ FlashAttention, https://github.com/Dao-AILab/flash-attention
6237
+
6238
+ BSD 3-Clause License
6239
+
6240
+ Copyright (c) 2022, the respective contributors, as shown by the AUTHORS file.
6241
+ All rights reserved.
6242
+
6243
+ Redistribution and use in source and binary forms, with or without
6244
+ modification, are permitted provided that the following conditions are met:
6245
+
6246
+ * Redistributions of source code must retain the above copyright notice, this
6247
+ list of conditions and the following disclaimer.
6248
+
6249
+ * Redistributions in binary form must reproduce the above copyright notice,
6250
+ this list of conditions and the following disclaimer in the documentation
6251
+ and/or other materials provided with the distribution.
6252
+
6253
+ * Neither the name of the copyright holder nor the names of its
6254
+ contributors may be used to endorse or promote products derived from
6255
+ this software without specific prior written permission.
6256
+
6257
+ THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
6258
+ AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
6259
+ IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
6260
+ DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
6261
+ FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
6262
+ DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
6263
+ SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
6264
+ CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
6265
+ OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
6266
+ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Binary file
metadata CHANGED
@@ -1,14 +1,14 @@
1
1
  --- !ruby/object:Gem::Specification
2
2
  name: onnxruntime
3
3
  version: !ruby/object:Gem::Version
4
- version: 0.7.7
4
+ version: 0.8.0
5
5
  platform: x86_64-darwin
6
6
  authors:
7
7
  - Andrew Kane
8
8
  autorequire:
9
9
  bindir: bin
10
10
  cert_chain: []
11
- date: 2023-07-25 00:00:00.000000000 Z
11
+ date: 2023-09-21 00:00:00.000000000 Z
12
12
  dependencies:
13
13
  - !ruby/object:Gem::Dependency
14
14
  name: ffi
@@ -55,7 +55,7 @@ required_ruby_version: !ruby/object:Gem::Requirement
55
55
  requirements:
56
56
  - - ">="
57
57
  - !ruby/object:Gem::Version
58
- version: '2.7'
58
+ version: '3'
59
59
  required_rubygems_version: !ruby/object:Gem::Requirement
60
60
  requirements:
61
61
  - - ">="