Add CNN vision support and image caption console #98
Conversation
This file contains hidden or 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
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Comment on lines
+361
to
+384
| if (modelMetaData.EncoderType != EncoderTypeEnums.VisionCNN) | ||
| { | ||
| Logger.WriteLine(Logger.Level.debug, $"Creating shared embeddings for both source side and target side. Shape = '({modelMetaData.SrcVocab.Count} ,{modelMetaData.EncoderEmbeddingDim})'"); | ||
| if (modelMetaData.SharedEmbeddings) | ||
| { | ||
| Logger.WriteLine(Logger.Level.debug, $"Creating shared embeddings for both source side and target side. Shape = '({modelMetaData.SrcVocab.Count} ,{modelMetaData.EncoderEmbeddingDim})'"); | ||
|
|
||
| srcEmbeddings = new MultiProcessorNetworkWrapper<IWeightTensor>(new WeightTensor(new long[2] { modelMetaData.SrcVocab.Count, modelMetaData.EncoderEmbeddingDim }, | ||
| raDeviceIds.GetNextItem(), initType: RandomInitType.Uniform, fanOut: true, name: "SharedEmbeddings", isTrainable: isSrcEmbeddingTrainable, learningRateFactor: encoderStartLearningRateFactor, dtype: elementType), DeviceIds); | ||
| srcEmbeddings = new MultiProcessorNetworkWrapper<IWeightTensor>(new WeightTensor(new long[2] { modelMetaData.SrcVocab.Count, modelMetaData.EncoderEmbeddingDim }, | ||
| raDeviceIds.GetNextItem(), initType: RandomInitType.Uniform, fanOut: true, name: "SharedEmbeddings", isTrainable: isSrcEmbeddingTrainable, learningRateFactor: encoderStartLearningRateFactor, dtype: elementType), DeviceIds); | ||
|
|
||
| tgtEmbeddings = null; | ||
| } | ||
| else | ||
| { | ||
| Logger.WriteLine(Logger.Level.debug, $"Creating embeddings for source side. Shape = '({modelMetaData.SrcVocab.Count} ,{modelMetaData.EncoderEmbeddingDim})'"); | ||
| tgtEmbeddings = null; | ||
| } | ||
| else | ||
| { | ||
| Logger.WriteLine(Logger.Level.debug, $"Creating embeddings for source side. Shape = '({modelMetaData.SrcVocab.Count} ,{modelMetaData.EncoderEmbeddingDim})'"); | ||
|
|
||
| srcEmbeddings = new MultiProcessorNetworkWrapper<IWeightTensor>(new WeightTensor(new long[2] { modelMetaData.SrcVocab.Count, modelMetaData.EncoderEmbeddingDim }, | ||
| raDeviceIds.GetNextItem(), initType: RandomInitType.Uniform, fanOut: true, name: "SrcEmbeddings", isTrainable: isSrcEmbeddingTrainable, learningRateFactor: encoderStartLearningRateFactor, dtype: elementType), DeviceIds); | ||
| srcEmbeddings = new MultiProcessorNetworkWrapper<IWeightTensor>(new WeightTensor(new long[2] { modelMetaData.SrcVocab.Count, modelMetaData.EncoderEmbeddingDim }, | ||
| raDeviceIds.GetNextItem(), initType: RandomInitType.Uniform, fanOut: true, name: "SrcEmbeddings", isTrainable: isSrcEmbeddingTrainable, learningRateFactor: encoderStartLearningRateFactor, dtype: elementType), DeviceIds); | ||
| } | ||
| } | ||
|
|
||
| Logger.WriteLine(Logger.Level.debug, $"Creating embeddings for target side. Shape = '({modelMetaData.TgtVocab.Count} ,{modelMetaData.DecoderEmbeddingDim})'"); | ||
| Logger.WriteLine(Logger.Level.debug, $"Creating embeddings for target side. Shape = '({modelMetaData.TgtVocab.Count} ,{modelMetaData.DecoderEmbeddingDim})'"); | ||
|
|
||
| tgtEmbeddings = new MultiProcessorNetworkWrapper<IWeightTensor>(new WeightTensor(new long[2] { modelMetaData.TgtVocab.Count, modelMetaData.DecoderEmbeddingDim }, | ||
| raDeviceIds.GetNextItem(), initType: RandomInitType.Uniform, fanOut: true, name: "TgtEmbeddings", isTrainable: isTgtEmbeddingTrainable, learningRateFactor: decoderStartLearningRateFactor, dtype: elementType), DeviceIds); | ||
| } | ||
| tgtEmbeddings ??= new MultiProcessorNetworkWrapper<IWeightTensor>(new WeightTensor(new long[2] { modelMetaData.TgtVocab.Count, modelMetaData.DecoderEmbeddingDim }, | ||
| raDeviceIds.GetNextItem(), initType: RandomInitType.Uniform, fanOut: true, name: modelMetaData.SharedEmbeddings ? "SharedEmbeddings" : "TgtEmbeddings", isTrainable: isTgtEmbeddingTrainable, learningRateFactor: decoderStartLearningRateFactor, dtype: elementType), DeviceIds); |
There was a problem hiding this comment.
When SharedEmbeddings is enabled the new logic still instantiates a target embedding wrapper (tgtEmbeddings ??= …), even though a shared source embedding has already been created. Both wrappers produce tensors named "SharedEmbeddings"; during SaveModel this results in two calls to model.AddWeights("SharedEmbeddings", …) which throws an ArgumentException for duplicate keys and prevents models with shared vocabularies from being saved. The previous implementation avoided this by not allocating a separate target embedding in the shared case. The additional allocation should be removed or redirected to reuse srcEmbeddings so only one set of shared weights is saved.
Useful? React with 👍 / 👎.
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.
Summary
Testing
dotnet build Seq2SeqSharp.sln -c Release(fails: dotnet CLI is not available in the execution environment)Codex Task