GoogleCloudPlatform / GoogleCloudPlatform/nodejs-docs-samples
Extract embeddings values using PredictionServiceClient
- Dominant language
- JavaScript
- Stars
- 3k
- Forks
- 2k
- Avg merge
- 2d 19h
- Merged PRs (30d)
- 44
Description
Hi! I'm trying to find how to extract the embeddings values to add them to a json files following this format following the example https://github.com/GoogleCloudPlatform/nodejs-docs-samples/blob/62253bb75d6068e1b44047eac5572246dfaa217b/ai-platform/snippets/predict-text-embeddings.js#L76C69-L76C69:
{"id":"0","embedding":[0.005268869,-0.019839672,0.0035084013,0.0536439,-0.028307509,0.0037563213, ...]}
But not sure how to access or put them into an array. The type is google.protobuf.IValue so in the example it iterates over an array google.protobuf.IValue[]. I'm doing:
```
const predictions = response.predictions;
for (const prediction of predictions) {
const value: google.protobuf.IValue = prediction;
console.log(`\t\tPrediction`, value.structValue); // structValue is the only one not returning undefined
// this returns: **fields: { embeddings: { structValue: [Object], kind: 'structValue' } }**
}
```
Thanks
Contributor guide
Research direction
Start with ai-platform/snippets/predict-text-embeddings.js at the referenced line and inspect how PredictionServiceClient responses are represented. Determine how the embeddings values should be accessed and whether the sample or its documentation needs clarification so the output matches the JSON format shown in the issue.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- google-cloud, javascript, node.js
- Domain
- api, machine-learning
- Issue type
- Documentation
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100