英文:
Why I can not train Matrix Factorization model with BigQuery even though I have purchased flex slots?
问题
我已经翻译好了您的内容，如下：
我已经与BigQuery纠缠了两个多星期。所有文档都只是在宣传该产品，没有提供任何帮助。
首先，我不想使用企业版或企业增强版。我目前正在评估BigQuery是否可以作为解决我的机器学习问题的可能解决方案。也就是说，我现在处于开发概念验证阶段。
如文档所述：
注意：矩阵分解模型仅适用于按固定费率计费的客户或具有预订的客户。建议按需客户使用灵活的插槽来进行矩阵分解。
也就是说，作为普通客户（不是购买企业版或企业增强版的客户），我必须购买灵活的插槽，然后进行预订设置吗？
我购买了100个灵活的插槽，只是想看看它如何运作，以及是否对我的情况有帮助。
然后，我从“创建预订”页面创建了一个预订。在“版本”下拉菜单中，我有3个选项：标准、企业和企业增强。
据我了解，“标准”是我可以用于我的概念验证的唯一选项，也就是说，我可以设置一个“标准”预订，运行SQL查询，看看模型是否适合我，然后删除预订。我没有选择“企业”或“企业增强”，因为它们按年计费，显然不符合我的要求。
创建了预订后，我运行了来自官方教程页面的SQL查询1：
#standardSQL
CREATE OR REPLACE MODEL bqml_tutorial.my_explicit_mf_model
OPTIONS
(model_type='matrix_factorization',
user_col='user_id',
item_col='item_id',
l2_reg=9.83,
num_factors=34) AS
SELECT
user_id,
item_id,
rating
FROM movielens.movielens_1m
这给了我以下错误：
“对于按需使用，不支持训练矩阵分解模型。要进行训练，请根据BigQuery公共文档中的说明设置预订（灵活或常规）。”
我做错了什么？我不应该购买会按年计费的东西，只是运行官方文档中的查询，对吗？我应该如何测试该产品，而不会陷入年度合同？我只想测试一个查询并仅支付一个查询的费用，而不是购买一年的合同？
英文:
I am strugguling for more than two weeks with BigQuery. All the docs are just ad ertizing the product, without any help.
First thing first, I dont want to use Enterprise Edition or Enterprise Plus Edition. I am currently evaluating BigQuery as a possible solution to my machine learning problem. That is, right now I am ate the stage of developing proof of conept.
As written on the docs:
> Note: Matrix factorization models are only available to flat-rate
> customers or customers with reservations. On-demand customers are
> encouraged to use flex slots to use matrix factorization.
That is, me as a regular customer (not one who purchased Enterprice or Enterprice Plus editions) have to by flex slots and then set up a reservation with it?
I bought 100 flex slots, with the idea just to see how it works and if it will be helpful for my case.
Then I created a reservation from the create reservation page. From the Edition dropdown I had 3 options: Standard, Enterprise and Enterprise Plus.
As far as I understand, Standard is the only option which I can use for my proof of a concept, that is, I can set up a Standard reservation, run the SQL query, see if the model works for me, and then delete the reservation. I did not choose Enerprice or Enterprise Plus because they are billed annually which obviously I dont want.
After I created the reservation, I run the SQL query which is from the official tutorial page
#standardSQL
CREATE OR REPLACE MODEL bqml_tutorial.my_explicit_mf_model
OPTIONS
(model_type='matrix_factorization',
user_col='user_id',
item_col='item_id',
l2_reg=9.83,
num_factors=34) AS
SELECT
user_id,
item_id,
rating
FROM movielens.movielens_1m
Which gives me the following error:
Training Matrix Factorization models is not available for on-demand usage. To train, please set up a reservation (flex or regular) based on instructions in BigQuery public docs.
What I am doing wrong? I am not supposed to buy something which will be billed annually so I just run a query from the official docs, am I? How am I supposed to test the product, without getting into an annual contract? I want just to test that one query and pay for that one query only, not to buy one year contract?
答案1
得分: 1
在创建预留后，允许您的项目使用该预留。通常，它包括创建插槽、预留和分配。
bq mk --reservation_assignment \
--reservation_id=${PROJECT}:${LOCATION}.${RESERVATION} \
--job_type=QUERY --assignee_type=PROJECT \
--assignee_id=${PROJECT}
有关更多信息，请参阅：使用 BigQuery Flex Slots 更高效地运行机器学习工作负载
英文:
After creating the reservation, allow your project to use the reservation.<br><br>Usually, it's create slots, reserve, assign.
bq mk --reservation_assignment \
--reservation_id=${PROJECT}:${LOCATION}.${RESERVATION} \
--job_type=QUERY --assignee_type=PROJECT \
--assignee_id=${PROJECT}
For more info: Using BigQuery Flex Slots to run machine learning workloads more efficiently<br>
通过集体智慧和协作来改善编程学习和解决问题的方式。致力于成为全球开发者共同参与的知识库，让每个人都能够通过互相帮助和分享经验来进步。
评论