continuedev/continue · error · Error

Unsupported embeddings type received: number[]

Error message

Unsupported embeddings type received: number[]

What it means

Companion check in getEmbedTexts: input is an array whose first element is neither an array nor a string — i.e. a flat array of numbers (token IDs, number[][]/number[] style pre-tokenized input). Bedrock embeddings require string inputs, so the adapter rejects it client-side with a clear message instead of sending an invalid AWS request.

Source

Thrown at packages/openai-adapters/src/apis/Bedrock.ts:622

    }
    const decoder = new TextDecoder();
    const decoded = decoder.decode(response.body);
    return JSON.parse(decoded);
  }

  private getEmbedTexts(body: EmbeddingCreateParams): string[] {
    const texts: string[] = [];
    if (typeof body.input === "string") {
      texts.push(body.input);
    } else if (body.input.length > 0) {
      const firstVal = body.input[0];
      if (Array.isArray(firstVal)) {
        throw new Error("Unsupported embeddings type received: number[][]");
      }
      if (typeof firstVal === "string") {
        texts.push(...(body.input as string[]));
      } else {
        throw new Error("Unsupported embeddings type received: number[]");
      }
    }
    return texts;
  }

  async embed(body: EmbeddingCreateParams): Promise<CreateEmbeddingResponse> {
    const texts = this.getEmbedTexts(body);

    let embeddings: number[][];
    if (body.model.startsWith("cohere")) {
      const payload = {
        texts,
        input_type: "search_document",
        truncate: "END",
      };
      const output = await this.getInvokeModelResponseBody(body.model, payload);
      embeddings = [output.embedding];
    } else if (body.model.startsWith("amazon.titan-embed")) {

View on GitHub (pinned to 5522c6f44c)

Solutions

1. Ensure every element of input is a string; cast/convert numbers via String(...) only if they were meant as text.
2. If the numbers are token IDs, decode to text with the originating tokenizer first.
3. Validate input shape before calling embed.

Example fix

// before
const res = await api.embed({ model: 'bedrock/titan-embed', input: [101, 2054] });

// after
const res = await api.embed({ model: 'bedrock/titan-embed', input: ['hello', 'world'] });

const ok = Array.isArray(body.input) ? body.input.every(x => typeof x === 'string') : typeof body.input === 'string';
if (!ok) throw new TypeError('input must be string or string[] for Bedrock');

function isStringArrayOrString(v: unknown): v is string | string[] {
  if (typeof v === 'string') return true;
  return Array.isArray(v) && v.length > 0 && v.every(x => typeof x === 'string');
}

When it happens

Trigger: Calling embed with input: [1, 2, 3] (flat number array of token IDs); mixing types like [42, 'text']; passing numeric IDs from an upstream tokenizer.

Common situations: Same as token-array scenarios: OpenAI-compatible code that sends token IDs, dynamic inputs coerced to numbers, or a JSON config supplying numbers where strings were intended.

Related errors

• Unsupported embeddings type received: number[][]
• Unsupported model: ${body.model}
• Query and chunks must not be empty
• Failed to fetch messages: ${response.statusText}
• Continue currently only supports text resources from MCP

AI-assisted analysis of continuedev/continue@5522c6f44c (2026-08-27). Data as JSON: /api/errors/63bb6d3daaa41171. Report an issue: GitHub.