Method handle_embeddings_impl

tools/server/server-context.cpp:4256–4352

}

std::unique_ptr<server_res_generator> server_routes::handle_embeddings_impl(const server_http_req & req, task_response_type res_type) {
    auto res = create_response();
    if (!params.embedding) {
        res->error(format_error_response("This server does not support embeddings. Start it with `--embeddings`", ERROR_TYPE_NOT_SUPPORTED));
        return res;
    }

    if (res_type != TASK_RESPONSE_TYPE_NONE && meta->pooling_type == LLAMA_POOLING_TYPE_NONE) {
        res->error(format_error_response("Pooling type 'none' is not OAI compatible. Please use a different pooling type", ERROR_TYPE_INVALID_REQUEST));
        return res;
    }

    const json body = json::parse(req.body);

    // for the shape of input/content, see tokenize_input_prompts()
    json prompt;
    if (body.count("input") != 0) {
        prompt = body.at("input");
    } else if (body.contains("content")) {
        res_type = TASK_RESPONSE_TYPE_NONE; // "content" field is not OAI compatible
        prompt = body.at("content");
    } else {
        res->error(format_error_response("\"input\" or \"content\" must be provided", ERROR_TYPE_INVALID_REQUEST));
        return res;
    }

    bool use_base64 = false;
    if (body.count("encoding_format") != 0) {
        const std::string & format = body.at("encoding_format");
        if (format == "base64") {
            use_base64 = true;
        } else if (format != "float") {
            res->error(format_error_response("The format to return the embeddings in. Can be either float or base64", ERROR_TYPE_INVALID_REQUEST));
            return res;
        }
    }

    auto tokenized_prompts = tokenize_input_prompts(ctx_server.vocab, ctx_server.mctx, prompt, true, true);
    for (const auto & tokens : tokenized_prompts) {
        // this check is necessary for models that do not add BOS token to the input
        if (tokens.empty()) {
            res->error(format_error_response("Input content cannot be empty", ERROR_TYPE_INVALID_REQUEST));
            return res;
        }
    }

    int embd_normalize = 2; // default to Euclidean/L2 norm
    if (body.count("embd_normalize") != 0) {
        embd_normalize = body.at("embd_normalize");
        if (meta->pooling_type == LLAMA_POOLING_TYPE_NONE) {
            SRV_DBG("embd_normalize is not supported by pooling type %d, ignoring it\n", meta->pooling_type);
        }
    }

    // create and queue the task
    json responses = json::array();
    auto & rd = res->rd;
    {