mitzen

Besides hooking up your search service to an AI, you might want to connect to your search service to retrieve data you already indexed. We will be creating index for our json data stored in a storage account via portal - it will automatically create index and indexer for us.  Besides JSON, you can index other data such as SQL database, mark down and even Apache Spark. Index allow our document to be ' structure and define for faster search '. Indexer perform actual indexing to our data and we can specify or schedule when we wants it. Pre-requisite  AI search  Storage acccount with hierarchical name enabled python code  First we click on " Import data " and select " Data Lake " And this is what our storage account looks like :- And then let's configure our json:- click " next " :- and then accept the default :- And then you can configure the schedule for our index to run. Use the default and configure it to run "Once". And finally provi...

One of the most overlook skills in cloud is designing or allocating subnet. Let's say your  vnet is created with address space of 10.0.0.0/16 - about 65,526 - we like to keep our VNET optimium with less fragmentations as possible.  Here is a list that we can use to ensure we are able to size our subnet for different setup /29, /28, /27 and /26. This make our life easier.  This would really depends on how your workload grows in a given subnet. For application that you would like to place them in a specific subnet and plan for certain expansion. If it grows beyond that, then you might need to provision a larger subnet. 

Cline provide free version of Ox Alpha too.  

We can use packages from a git repository, for example -  https://github.com/kepungnzai/dbt-tutorial-setup.git To do that we can update our packages.yml in the root folder - packages.yml  packages :   - package : dbt-labs/dbt_utils     version : ">=1.3.0"   - package : dbt-labs/audit_helper     version : 0.14.0   - git : https://github.com/kepungnzai/dbt-tutorial-setup.git     revision : main then run  dbt deps And dbt get the packages from git And as you can see, the package apear under our folder - dbt_packages:-  

In dbt we can check and assess our data to see if it is stale or has not been updated for some time.  We can use the following configuration here to do it - config-> freshness. # dbt test --select "test_type:data" version : 2 sources :   - name : SNOWFLAKE_SAMPLE_DATA     database : SNOWFLAKE_SAMPLE_DATA       schema : TPCDS_SF10TCL     config :       freshness :         error_after :           count : 2           period : day     tables :       - name : store_sales We can run " dbt source freshness ".  In the first section, it error out because the day last update is 1 year 8 moths ago  We can easily bump up the number here to 3285 (9 years).  # dbt test --select "test_type:data" version : 2 sources :   - name : SNOWFLAKE_SAMPLE_DATA     database : SNOWFLAKE_SAMPLE_DATA       schema :...

In dbt we can easily create macro to automate common use tasks. One of that would be a audit log operation.  In  your folder called macro, please create a file called "audit_operation.sql". {% macro audit_operation(activity_type) %}   insert into raw . audit (run_id, TYPE_ACITIVITY) values ( '{{ 1 }}' , '{{ activity_type }}' ) {% endmacro %} And then we can use it in our dbt project like so. Instead of placing the SQL code in the commented code, we use macro.  name : jaffle_shop profile : jaffle_shop seed-paths : [ "seeds" ] model-paths : [ "models" ] macro-paths : [ "macros" ] test-paths : [ "tests" ]           # ← Make sure this exists clean-targets :   - "target"   - "dbt_packages" seeds :   jaffle_shop :     +schema : raw # on-run-start:  "insert into raw.audit (run_id, TYPE_ACITIVITY) values ('1', 'pre-hook')" on-run-start : "{{ audit_operation('macro-pre-hoo...

 dbt uses jinja as a template engine to work with file. Therefor we can configure our variable 2 ways. 1. dbt_project.yaml or vars.yml (on the project root) - we can introduce variable by using the followings - as you can see here we are introducing a variable call person.  name : jaffle_shop profile : jaffle_shop seed-paths : [ "seeds" ] model-paths : [ "models" ] macro-paths : [ "macros" ] test-paths : [ "tests" ]           # ← Make sure this exists clean-targets :   - "target"   - "dbt_packages" seeds :   # Builds seeds into '<your_schema_name>_raw'   jaffle_shop :     +schema : raw vars :   person : JohnDoe this is what your vars.yml looks like  vars :   person : JohnDoe2   sold_date_sk : 20160901 2. passing variable in from the command line using --var argument. Please note variable are passed in as json - so we need to do something like this. Please note: it is important to use double quote ...

This is quite a good place to get started with enterprising your dbt   https://www.phdata.io/blog/accelerating-and-scaling-dbt-for-the-enterprise/

We can extend our dbt capabilities through packages just like packages in other languages like nuget in dotnet. In fact, we might already been using packages already, if you are following previous tutorials where we use materialization and sources keyword - this package is called dbt_utils.  Let's see how we can confugure it :- packages :   - package : dbt-labs/dbt_utils     version : " >=1.3.0 "   - package : dbt-labs/audit_helper     version : 0.14.0 And them run " dbt deps " That's how we can install it If you're looking out for more dependencies then please go to dbt hub.https://hub.getdbt.com/ So you have the dbt util install how can you use it? Let's try    dbt_utils.equal_rowcount - this is quite a simple example to compare 2 table in your test if they have the same row count. In your model directory and look for your model that you like to compare. In my case it is model/snowflake_sample_data and then add the tests in  ...

We can use python in our dbt project as well. Here is an example of code that you can place under model folder def model ( dbt , session ):     dbt . config ( materialized = ' table ' )         source_store_sales = dbt . source ( ' SNOWFLAKE_SAMPLE_DATA ' , ' store_sales ' )         return source_store_sales This is equilvalent to : {{ config( materialized = ' view ' )}} with source_store_sales as (     select * from {{ source ( ' SNOWFLAKE_SAMPLE_DATA ' , ' store_sales ' ) }} ), renamed as (     select       *     from source_store_sales ) select * from renamed There are limitation when using python code where it only supports table and incremental. 

dbt data test is used to test validity of the data during the Transform and load stage and that can be easily setup using the followings example Please note: The model is your file name which is  " snowflake_sample_data_store_sales ". Remember dbt always use the filename as model name.  Data test models :   - name : snowflake_sample_data_store_sales     description : List of store sales records with basic cleaning and transformation applied.     columns :       - name : SS_ITEM_SK         description : The unique key for each location.         data_tests :           - not_null if you run dbt test now, you will get the following output here: Unit test  We also have unit test in dbt - that allow us to run unit test against our data. This is useful to ensure our sql used for our TL is working as expected.  You need to ensure the model name matches your filenam...

We can find the actual path of your executable by using powershell command. Normal we added some exec somewhere and forgot about it.  The powershell command can help to trace where it is, save me bunch of time trying to look for this exec. ( Get-Command dbt ) .Source 

Let's say you do not want certain resources to be created. This can be controlled via Azure policy and can be scope to subscription or resource group level. We will prevent storage account from being created in a resource group and here is how we can do it. First goto Azure Policy - Authoring - Definition and look for ' Not allowed resource types '.  Then click on "Assign Policy" and for my test, I will scope this to Resource group level. And then provide a name to it like so, Next, is where we specify "Resource Type" - let's select storageAccounts and then click on "Next".  And now if you switch over to "Assignments", you will notice your policy gets created.  Now it is time for test out your policy by creating a storage account in myfdrydev-rg . And then you will be hit with an error. This is what our policy definition looks like in json {   "properties" : {     "displayName" : "Not allowed resource type...

To kick my very own LLM model that can be train and learned from simple dataset - i started off with a bare minimum model that allows me to learn really simple stuff like 1 + 1 = 2, 2 + 2 = 4.   So first we outline our vocabulary and size - how we are representing this information to the LLM model Vocabulary and token Then we tokenize those input for training and inference. This model only understands these vocab Special: <pad> , <start> , <eos> Operators: + , - , = Numbers: 1 , 2 , 3 , 4 Simple transformer with: Embedding Layer : Converts token IDs to 32-dim vectors and this is the layer where we handle our vocab      # Token embedding   self . embedding = nn . Embedding ( vocab_size , d_model , padding_idx = 0 ) Positional Encoding : Learnable position embeddings      # Positional encoding (learnable)   self . pos_embedding = nn . Embedding ( seq_len , d_model ) Transformer Encoder : 1 layer with 2 attentio...