Add GCP VertexAI adapter

[akihikokuroda]

Which issue(s) does this pull-request address?

Ref: #141

Description

So far _stream in chat.ts has been tried. The other 3 _genereate and _stream methods have not been tried yet.

Here are some links that I've been looking:

https://www.npmjs.com/package/@google-cloud/vertexai?activeTab=readme#get-nonstreamed-text-responses
https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/inference
https://github.com/googleapis/nodejs-vertexai/blob/main/src/models/generative_models.ts

Checklist

• I have read the contributor guide
• Linting passes: yarn lint or yarn lint:fix
• Formatting is applied: yarn format or yarn format:fix
• Unit tests pass: yarn test:unit
• E2E tests pass: yarn test:e2e
• Tests are included
• Documentation is changed or added
• Commit messages and PR title follow conventional commits

[Tomas2D]

Good work so far!

[Tomas2D]

Great job! I just noticed that you did not add support for Constraint Decoding. From the documentation it seems that VertexAI supports JSON Schema.

Please look at how it is done in BAM Adapter (

bee-agent-framework/src/adapters/ibm-vllm/llm.ts

Lines 225 to 249 in d82be6e

	protected _prepareParameters(overrides?: GenerateOptions): typeof this.parameters {
	const guided = overrides?.guided ? {} : (this.parameters.decoding ?? {});
	const guidedOverride = overrides?.guided;
	if (guidedOverride?.choice) {
	guided.choice = { ...guided.choice, choices: guidedOverride.choice };
	} else if (guidedOverride?.grammar) {
	guided.grammar = guidedOverride.grammar;
	} else if (guidedOverride?.json) {
	guided.json_schema = isString(guidedOverride.json)
	? JSON.parse(guidedOverride.json)
	: guidedOverride.json;
	} else if (guidedOverride?.regex) {
	guided.regex = guidedOverride.regex;
	} else if (!isEmpty(guidedOverride ?? {})) {
	throw new NotImplementedError(
	`Following types ${Object.keys(overrides!.guided!).join(",")}" for the constraint decoding are not supported!`,
	);
	}
	return {
	...this.parameters,
	decoding: guided,
	};
	}

).

[Tomas2D]

Great job! 🚀