Read & perform inference on networks for which the hidden-layer width varies across layers #166
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This commit switches the metadata_ component of the trainable_engine_t type from being an array of string_t objects to being an instance of the new metadata_t type.
This commit switches the metadata component of the exchange_t type from being an array of string_t objects to being an instance of the new metadata_t type.
This commit switches the metadata component of the inference_engine_t type from being an array of string_t objects to being an instance of the new metadata_t type.
This commit edits the JSON file readers and writers to
1. Switch keys "tensor_range" -> {inputs_range|outputs_range}
2. Allow flexible line-position of tensor ranges and hidden layers.
This commit edits the inference_engine_t from_json constructor to allow flexible line-position of the metadata object.
This commit switches from specifying compile-time constants for the number of lines occupied by metadata and tensor_range JSON objects to computing the number of lines at runtime in inference_engine_t's to_json type-bound procedure. This will make to_json adaptable in case the size of these objects changes in the future.
This commit switches from specifying a compile-time constant for the number of lines occupied by neuron JSON objects now the number of lines per neuron is computed at runtime using by checking the array size of the result of invoking the to_json type-bound procedure on a neuron_t prototype.
This commit adds example/read-query-infer.f90, which works with neural nets for which the hidden-layer width varies from one layer to the next.
rouson
added a commit
that referenced
this pull request
Jul 4, 2024
With this commit, example/read-query-infer.f90 returns outputs identital to the inputs for the full range of inputs using the uneven.json file attached topull request #166 on https://github.com/berkeleylab/inference-engine.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
With this pull request, executing
gives the desired result of producing outputs identically equal to the corresponding inputs on a network with varying hidden-layer widths such as the attached file uneven.json.gz when uncompressed. The new
example/read-and-query.f90program generates random inputs. An example execution of the above line produces the output