summaryrefslogtreecommitdiffstats
path: root/model-integration
Commit message (Collapse)AuthorAgeFilesLines
* Revert "Deopend on component to get AbstractComponent"Bjørn Christian Seime2023-02-281-6/+0
|
* Deopend on component to get AbstractComponentJon Bratseth2023-02-281-0/+6
|
* Extend AbstractComponent not AbstractResourceBjørn Christian Seime2023-02-283-9/+9
|
* Replace `OnnxEvaluatorCache` with OnnxRuntimeBjørn Christian Seime2023-02-2714-228/+325
| | | | | | Require an `OnnxRuntime` instance to create `OnnxEvaluator` instances. Cache underlying `OrtSession` instead of `OnnxEvaluator`. Move static helpers for checking Onnx runtime availability from `OnnxEvaluator` to `OnnxRuntime`.
* handle non-identifier onnx input/output names: instead of the conflictingArne Juul2023-02-221-0/+30
| | | | | ad-hoc code in OnnxEvaluator, do it as part of general input/output mapping in OnnxModel.
* Cache Onnx model instancesBjørn Christian Seime2023-02-223-0/+141
| | | | | | Manage lifecycle of OnnxEvaluator instances explicitly to allow instances to be cached without use WeakHashmap/finalizers. Inject shared Onnx model cache in ModelsEvaluator.
* Make OnnxEvaluator closableBjørn Christian Seime2023-02-221-1/+12
|
* Implement equals()/hashCode()Bjørn Christian Seime2023-02-221-1/+17
| | | | Required for using instances as key in HashMap
* Add initial text generator componentLester Solbakken2023-02-189-2/+410
|
* ensure outputs with names as promised by getOutputInfo()Arne Juul2023-02-103-1/+123
|
* Fix GPU detectionMartin Polden2023-02-091-1/+1
|
* Allow fallback to CPU if nodes are provisioned without GPUMartin Polden2023-02-082-13/+24
|
* Remove 'required' attributeMartin Polden2023-01-262-22/+14
|
* Skip CUDA entirely if GPU device is optionalMartin Polden2023-01-262-24/+39
| | | | CUDA may fail after its library is loaded, e.g. when the session is created.
* Do not filter on error codeMartin Polden2023-01-241-4/+1
|
* Support configuration of GPU device to use in ONNX modelMartin Polden2023-01-231-1/+24
|
* Add JNA in lib/jars + export as packages from jdisc_coreBjørn Christian Seime2023-01-201-1/+21
| | | | Jdisc-core will embed JNA. The JNA in lib/jars is used by fatjars only.
* Unify on Streams.toList()Henning Baldersheim2023-01-172-4/+3
|
* Exclude in fat-depsJon Bratseth2023-01-091-1/+4
|
* Huggingface not needed in config modelJon Bratseth2023-01-091-0/+1
|
* Allow model-integration:huggingface dependenciesJon Bratseth2023-01-061-0/+6
|
* DJL-based HuggingFaceEmbedder prototypeAndrii Yurkiv2023-01-044-0/+232
|
* trying to make an OnnxEvaluator with empty path would failArne Juul2022-12-121-0/+8
|
* Merge pull request #25075 from vespa-engine/arnej/onnxruntime-as-bundleArne H Juul2022-12-051-3/+5
|\ | | | | add container-onnxruntime-bundle
| * rename to just "container-onnxruntime"Arne Juul2022-12-021-1/+1
| |
| * add container-onnxruntime-bundleArne Juul2022-12-021-3/+5
| |
* | Revert "Revert collect(Collectors.toList())"Henning Baldersheim2022-12-044-4/+4
| |
* | Revert collect(Collectors.toList())Henning Baldersheim2022-12-044-4/+4
| |
* | collect(Collectors.toList()) -> toList()Henning Baldersheim2022-12-024-4/+4
|/
* Revert "Revert "- Reduce usage of guava.""Henning Baldersheim2022-12-013-28/+29
|
* Revert "- Reduce usage of guava."Henning Baldersheim2022-12-013-29/+28
|
* - Reduce usage of guava.Henning Baldersheim2022-12-013-28/+29
| | | | - Ensure that tests relying on order are determinsitic.
* Revert "Revert "Revert "Revert "Balder/model importing code in config model ↵Henning Baldersheim2022-11-095-14/+36
| | | | [run-systemtest]""""
* Revert "Revert "Revert "Balder/model importing code in config model ↵Henning Baldersheim2022-11-095-36/+14
| | | | [run-systemtest]"""
* Revert "Revert "Balder/model importing code in config model [run-systemtest]""Henning Baldersheim2022-11-095-14/+36
|
* Revert "Balder/model importing code in config model [run-systemtest]"Henning Baldersheim2022-11-075-36/+14
|
* Revert "Revert "Balder/model importing code in config model [run-systemtest]""Henning Baldersheim2022-11-045-14/+36
|
* Revert "Balder/model importing code in config model [run-systemtest]"Henning Baldersheim2022-11-045-36/+14
|
* Avoid deprecation warnings.Henning Baldersheim2022-11-041-0/+2
|
* Leave the existing interface as is and add 2 accessors for accessing the ↵Henning Baldersheim2022-11-045-17/+36
| | | | tensors directly.
* Avoid converting tensor to strings and back again.Henning Baldersheim2022-11-045-12/+13
|
* Revert "Since large constants are always tensors avoid creating a string ↵Harald Musum2022-11-045-13/+12
| | | | just…"
* Handle small and large Constants in a uniform way.Henning Baldersheim2022-11-034-5/+5
|
* Since large constants are always tensors avoid creating a string just to ↵Henning Baldersheim2022-11-035-7/+8
| | | | | | convert it back to tensors. That is a lot of work for nothing.
* Remove unnused vespa-convert-tf2onnx script.Arnstein Ressem2022-10-272-104/+1
|
* Update ABI spec format, and update all specsjonmv2022-10-251-1/+1
|
* support models without tokenTypeIds (like DistilBERT)Christophe Jolif2022-10-204-3/+97
|
* Use tensor vocabularyJon Bratseth2022-09-287-16/+16
|
* Drop 'arithmetic' from nameJon Bratseth2022-09-287-36/+34
|
* Merge pull request #24058 from vespa-engine/bratseth/mode-def-fileJon Bratseth2022-09-153-3/+32
|\ | | | | Move the BERT config def to model-integration