تعرف على خياراتك قبل أن تصبح واحدًا من مبدعي تابلو

In the last few years, the KSA market has witnessed a definitive shift to self-service consumption tools anchored in data democratization. The transition is of paramount importance for products that can scale at an enterprise level, AI-ML products, and those that can address data governance and quality management issues.

The shift is also significant for enterprise transformation catalysts that take an ecosystem approach with proven expertise in developing and executing comprehensive and unified data strategies, data engineering and data governance paradigms.

Thus the time is ripe for an innovation-led, experience-driven enterprise like Beinex to spearhead Digital and Analytics Transformations in KSA.

[sc name="quote" quote="“Beinex is pleased to formalize its presence in the KSA market by opening an Office in Riyadh. We, as an enterprise, are 100% aligned with Vision 2030 as put forth by the KSA and see tremendous value getting unlocked as the vision is realized. We look forward to expanding our footprint in the domains of Artificial Intelligence, Sustainability, Digital Transformation, Analytics and allied areas. The Kingdom envisions itself to be at the forefront of data and artificial intelligence-based economies, and Beinex is committed to playing its part in supporting and fulfilling this vision,”" author="Indumon Das, Founder and Managing Director of Beinex,
marking the occasion of the office’s opening, noted."][/sc]

Middle East Banking AI & Analytics Summit

Beinex is super excited to be a part of the 6th Middle East Banking AI & Analytics Summit on May 10, 2023. With the motto, "Accelerating Innovation in Banking with AI and Analytics Strategies", the summit aims to revolutionise the financial and banking space in KSA using AI. We are ready to witness and participate in panel discussions, fireside chats, keynote presentations, roundtable discussions, and conversational Q&A sessions with thought leaders on exploiting the Power of AI and Analytics for a futuristic banking ecosystem.

Middle East Enterprise AI & Analytics Summit

Also, we are enthusiastic to participate in the Middle East Enterprise Al and Analytics Summit on May 11, 2023. Its vision is to curate a world-class platform for tech leaders in the region to connect, communicate and collaborate under the theme "Accelerating Innovation in Enterprises with Applied Al and Analytics Strategies". Beinex is looking forward to connecting with thought leaders and high-level decision-makers in Al, and Data Analytics at #MEEAI 2023 to participate in discussions and to be a part of the transformation journey.

The Power of Beinex

Beinex drives a cohesive, unified digital ecosystem to help customers address their needs, assess products and operations, understand market requirements and evaluate overall business performance.

It is a multinational firm exploring the endless possibilities of data for Cloud, Analytics, Artificial Intelligence, Machine Learning, and Automation. In effect, Beinex architects, guides, leads, and implements solutions in Analytics, AI, and ML for the spheres of Digital Transformation, GRC, and Risk & Audit Transformation.

Partnerships make Beinex stronger. The company has solid partnerships with some of the leading technology firms, research labs, and universities around the globe. Businesses can leverage the power of the Beinex partner ecosystem to maximize the value of their end-to-end analytics journey.

Beinex Digital, a part of Beinex Holdings, is a digital transformation entity with a comprehensive suite of independent products focused on addressing specific business gaps, use cases, and needs. It incorporates a spectrum of solutions in the domains of Employee Health, Safety and Environment, Enterprise Product Management and Enterprise Performance Management.

Beinex is also the product champion for Aurex – Augmented Risk and Audit Analytics – a unique single-platform solution for Integrated Risk Management, Governance, Audit, Compliance, BCM, and Analytics functions. It is the first-of-its-kind product that streamlines risk and audit verticals for enterprises worldwide and is a Unified Digital Assurance Ecosystem.

Present in three continents, Beinex enables its clients to analyze data, mitigate risks, identify opportunities and automate processes.

Beinex Office Address (KSA):

Beinex Advanced Information Technology
3141, Anas Bin Malik,
8292 Al Malqa Dist
P. O. Box 13521,
Riyadh, Kingdom of Saudi Arabia
Email: Info@beinex.com

AWS Systems Manager seamlessly operates with both Windows and Linux OS and integrates with CloudWatch metrics, CloudWatch Dashboard, and AWS Config. Moreover, it enables the creation of resource groups spanning various AWS services, allowing for aggregated operational data viewing and facilitating monitoring, troubleshooting, and resource group-specific actions.

Common use cases and best practices for AWS Systems Manager capabilities are listed below:

Automation

Create self-service Automation runbooks for infrastructure.

Leverage AWS Systems Manager's Automation feature to simplify the process of generating Amazon Machine Images (AMIs) from the AWS Marketplace or custom AMIs. This can be achieved using public Systems Manager documents (SSM documents) or by creating custom workflows tailored to your specific needs.

Two recommended Automation runbooks for building and maintaining AMIs are AWS-UpdateLinuxAmi and AWS-UpdateWindowsAmi.

Inventory

Audit Application Configurations Over Time with Inventory

Use AWS Systems Manager's Inventory in conjunction with AWS Config to conduct comprehensive audits of your application configurations as they evolve over time.

Maintenance Windows

Scheduled Maintenance for Node Management

Define a routine schedule for executing potentially disruptive actions on your nodes, including activities like operating system (OS) patching, driver updates, or software installations.

For guidance on choosing between State Manager and Maintenance Windows, both key capabilities of AWS Systems Manager, refer to "Choosing between State Manager and Maintenance Windows."

Parameter Store

Centralised Management of Global Configuration Settings

Employ Parameter Store, a pivotal feature within AWS Systems Manager, to centrally manage and maintain global configuration settings.

Patch Manager

Efficient Patch Rollouts and Enhanced Fleet Compliance

Make use of Patch Manager, an integral component of AWS Systems Manager, to orchestrate the deployment of patches across your fleet of instances while also gaining enhanced visibility into fleet compliance.

To bolster your security posture, integrate Patch Manager with AWS Security Hub for real-time compliance alerts and monitoring of patching status.

Optimise Patch Compliance Scans

It is advisable to employ only one method at a time for scanning managed nodes for patch compliance to prevent accidental overwriting of compliance data.

Run Command

Effortless Instance Management without SSH Access

Harness the power of EC2 Run Command within AWS Systems Manager for managing instances at scale without the need for SSH access.

Secure and Controlled Command Execution

Audit all API calls related to Run Command by utilising AWS CloudTrail. To maintain data security, avoid including sensitive information in plaintext when sending commands via Run Command. Utilise SecureString parameters for encryption of sensitive data used in Systems Manager operations.

Staged Command Execution and Access Control

Employ the targets and rate control features in Run Command for executing staged command operations and ensure fine-grained access permissions through AWS Identity and Access Management (IAM) policies.

Session Manager

Comprehensive Session Activity Auditing

Audit session activity within your AWS account by making use of AWS CloudTrail.

Session Data Logging and Access Control

Log session data in your AWS account via Amazon CloudWatch Logs or Amazon S3, enabling tighter control over user session access to instances. Additionally, you can manage access to commands within a session and customise SSM-user account administrative permissions.

State Manager

Regular SSM Agent Updates and Tags for Node Management

Ensure that SSM Agent is updated at least once a month by utilising the pre-configured AWS-UpdateSSMAgent document. For Windows instances, upload PowerShell or DSC modules to Amazon S3 and use AWS-InstallPowerShellModule.

Employ tags to create application groups for your nodes and target nodes using the Targets parameter instead of specifying individual node IDs. Automate the remediation of findings generated by Amazon Inspector using Systems Manager.

Centralized Configuration Repository and Document Sharing

Utilize a centralised configuration repository for your SSM documents and share these documents across your organisation for streamlined management.

Managed Nodes

Time Accuracy for Optimal Operations

Accurate time references are essential for Systems Manager to perform operations effectively. Ensure that your node's date and time settings are correct, as discrepancies may lead to errors or incomplete functionality, and can result in nodes not being included in your managed nodes lists.

Case Study: A Global Cloud Solutions and Services Company Enhances Scalability and Efficiency with AWS Systems Manager

Client: A Global Cloud Solutions and Services Company

A technology services company that specialises in helping organisations across 120 countries adopt modern technologies and manage them efficiently. They focus on creating solutions for hybrid and multi-cloud environments.

Requirement: Finding Scalability on AWS Systems Manager

The client faced a significant challenge in managing multi-cloud environments at scale reliably and cost-effectively. Manually handling activities across hundreds of thousands of different compute instances was resource-intensive and delayed issue resolution. They needed a solution that could run both on-premises and on the cloud and wanted a single tool for managing their suite of solutions.>
Challenges

Manual management of hundreds of thousands of compute instances.

Delays in issue resolution due to human intervention.

Need for a solution that could operate on-premises and in the cloud.

Desire for a unified tool to manage their suite of solutions.

Process: Supporting Automation, Staff Productivity, and Transparency on AWS

The client began using AWS Systems Manager in 2015 for various products and extended its use to other cloud environments in 2019. Since 2019, the client has utilised AWS Systems Manager to power patching activities across all major cloud providers they support. They perform mass patching at scale, covering over 62,000 VMs across all their managed services. VM Management automates traditionally manual tasks like patching, agent distribution, server diagnostics, and issue remediation. It significantly reduces labour, costs, and errors associated with manual tasks, enhancing security and efficiency.

SmartTickets, a component in VM Management, handled thousands of incidents and automated responses using AWS Systems Manager, saving time and reducing costs for the company. They also used Amazon CloudWatch for monitoring and observability and automated runbooks for real-time monitoring and alerts.

AWS Systems Manager provides a single-pane view of environments, improving customer visibility and decision-making.

Result: Taking Automation to the Next Level on AWS

The client plans to develop custom runbooks with customers and further automate responses and resolutions using AWS Systems Manager. They have successfully solved industry challenges by saving time, cutting costs, and reducing complexity for both their customers and themselves.

Key Takeaway

The client leveraged AWS Systems Manager to streamline and automate their operations, resulting in improved efficiency, cost reduction, and enhanced customer satisfaction. With automation, they can swiftly respond to and resolve issues, meeting customer expectations effectively.

How Beinex Can Help You

Beinex is an AWS consulting partner, and we empower customers to host their BI solutions and much more on the cloud. Our cloud migration experts bring in best-in-class stability and reliability by understanding your business strategy and working closely with you to deploy AWS infrastructure as a service.

Enterprises worldwide have perceived the potential benefits of AI for their operations. AI gives humans the freedom to make insightful decisions while allowing a computer to perform other preset tasks that necessitates the development of such technologies in the first place. These tools assist you in developing, but they also aid in optimising networks and workflows.

A list of Artificial Intelligence tools is given below:

1. Scikit Learn
2. Tensorflow
3. Theano
4. Caffe
5. MxNet
6. Keras
7. PyTorch
8. CNTK

SCIKIT Learn

Known to be the most wanted tool in the library of Machine Learning for the python programming language, Scikit learn offers a wide range of tools for statistical modelling, Predictive analytics and very many other machine learning tasks. It underpins many administered and unsupervised learning calculations. It is a perfect tool for fledgling, and it incorporates direct and calculated relapses, choice trees, bunching, k-implies, etc.

Tensorflow

TensorFlow is an end-to-end open-source platform with a flexible ecosystem of tools for creating Machine Learning applications. It allows Google's voice-recognition tool to spot queries in photos and understand audibly stated phrases.

Theano

Theano was created to simplify and speed up the creation of sophisticated learning models so that they might be used in creative projects. It's written in Python and can run on both GPUs and CPUs. It generates elevated information counts that are often higher than when it runs solely on the CPU. Theano's speed makes it highly cost-effective to perform any complex calculations.

Caffe

The Berkeley Vision and Learning Center (BVLC) and network donors collaborated to construct Caffe, a deep learning structure that prioritises articulation, speed, and assessed quality. Google's Deep Dream uses Caffe Framework. CAFFE is a Python-interfaced BSD-authorized C++ library.

MxNET

MxNET uses a 'forgetful back prop' to barter computation time for memory, which is highly useful for recurrent nets on very long sequences. As it is an easy-to-use support for multi-GPU and multi-machine training, scalability is a priority during the design process. There are a lot of intriguing features, such as the ability to write custom layers in high-level languages. Unlike almost all other significant frameworks, it is not explicitly regulated by a vast corporation, which is suitable for an open-source, community-developed framework.

Keras

Keras is what you need if you like Python and how it works. It is a high-end library that tackles neural networks highly effectively for recurrent nets on very long sequences, which it achieves by utilising Theano and TensorFlow in the backend. It recognises the architecture that relates to specific issues. It aids in the detection of problems by using photos with weights. It optimises the results of a network by configuring it. Keras provides an abstract structure that can be transformed into any other framework for compatibility or performance.

Pytorch

The code for Pytorch, a Facebook-created artificial system, is easily accessible on Github. There are over 22000 stars on it. The framework has been in high demand in recent years, and it is still being developed. PyTorch uses reverse-mode auto-differentiation to modify network behaviour arbitrarily with zero lag or overhead, speeding up research iterations. Its deep learning framework is optimised for achieving state-of-the-art results in research.

CNTK

The Microsoft Cognitive Toolkit (CNTK) is an open-source, unified toolkit that describes neural networks as computational steps via a directed graph. Users utilise CNTK to release and merge popular types of models, such as DNNs, CNNs, RNNs, and LSTMs. It employs stochastic gradient descent (SGD), which learns through parallelisation and automatic differentiation across multiple servers and GPUs. Because of its open-source licenses, anyone can try out CNTK

Machine Learning Tools

Machine learning tools are algorithmic applications of artificial intelligence that allow systems to learn and develop without human input; data mining and predictive modelling are similar concepts. They will enable the software to improve its accuracy in anticipating outcomes without programming it directly. Some top Machine Learning Tools are enlisted below:

1. Microsoft Azure Machine Learning
2. IBM Watson
3. Google TensorFlow
4. Amazon Machine Learning
5. OpenNMS
6. Google Colab
7. Apache Mahout
8. Shogun

Microsoft Azure Machine Learning

Microsoft Azure Machine Learning is a cloud platform for building, training, and deploying AI models. Microsoft is constantly updating and improving its machine learning tools, and it just announced changes to Azure Machine Learning, including the retirement of the Azure Machine Learning Workbench.

IBM Watson

Watson Machine Learning is a cloud service from IBM that leverages data to deploy machine learning and deep learning models. Users can use this machine learning application to execute two basic machine learning operations: training and scoring. Remember that IBM Watson is best suited for developing machine learning applications via API connections.

Google TensorFlow

TensorFlow is an open-source software library for dataflow programming that Google uses for research and production. TensorFlow is, at its core, a machine learning framework. This machine learning tool is new to the market and is rapidly evolving. The ease with which TensorFlow allows developers to visualise neural networks is perhaps the most appealing feature.

Amazon Machine Learning

Amazon Machine Learning is used for creating and predicting Machine Learning models. Amazon Machine Learning comes with an automatic data transformation tool, which makes the machine learning tool even more user-friendly. Amazon also offers other machine learning tools, such as Amazon SageMaker, a fully-managed platform that makes using machine learning models simple for developers and data scientists.

OpenNMS

Open Neural Networks Package is a neural network implementation software library. OpenNMS, written in the C++ programming language, allows you to download its whole library from GitHub or SourceForge.

Google Colab

Google Colab is a cloud service supported by Python. It will assist in developing machine learning applications using PyTorch, Keras, TensorFlow, and OpenCV libraries. It facilitates machine learning and is accessible through Google Drive.

Apache Mahout

Apache Mahout is an Apache Software Foundation project that employs the MapReduce paradigm and is built on top of Apache Hadoop. It's also utilised to construct scalable, distributed machine learning algorithms for clustering, collaborative filtering, and classification. Mahout includes Java libraries for popular math algorithms and operations and foundational Java collections, concentrating on statistics and linear algebra.

Shogun

Shogun is an open-source ML platform, an open-source machine learning software library built in C++. It employs a diverse set of unified and efficient machine learning techniques. Shogun provides a well-organised implementation of all standard machine learning methods and is a critical player in ML education and development.

Robotic Process Automation Tools

Robotic Process Automation (RPA) tools are commonly used for task automation configuration. These tools are essential for automating repetitive back-office activities. With RPA Tools, we acquire a virtual employee who can execute repetitive tasks efficiently and, at less cost, than humans.

The following is a curated list of the top RPA tools:

1. Keysight's Eggplant
2. Inflectra Rapise
3. Blue Prism
4. UiPath
5. Automation Anywhere
6. Pega
7. Contextor
8. Nice Systems

Keysight's Eggplant

Eggplant RPA is a solution designed for process experts to automate the execution of repetitive tasks. It is compatible with apps such as SAP, Oracle, etc. and provides increased productivity and reduces errors.

Inflectra Rapise

Rapise by Inflectraina, a test automation solution, is in its seventh iteration and specialises in complicated applications like MS Dynamics, Salesforce, and SAP. Rapise now can automate Web, Desktop, and Mobile apps and supports hybrid business settings.

Blue Prism

Blue Prism RPA supports all core capabilities and is used with any application on any platform. You will need programming abilities to utilise this application, but it is user-friendly for developers. Blue Prism is ideal for medium and large businesses.

UiPath

UiPath is a user-friendly system that delivers security by handling credentials, encrypting data, and controlling access based on role. It is an open platform, adaptable for any business size and capable of handling complex procedures.

Automation Anywhere

Automation Anywhere provides core functions and security through authentication, encryption, and credentials. It is an easy-to-use solution ideal for medium and big businesses that offers both on-premise and cloud-based services.

Pega

Pega is a business process management platform that is hosted in the cloud. This is ideal for medium and large organisations and solely delivers cloud-based solutions or services. Pega is compatible with Windows, Linux, and Mac and can be installed on desktop servers.

Contextor

Contextor is an excellent fit for any size front office and works with all workstation applications. It supports Citrix and RDP hybrid virtualisation environments and provides on-premise and cloud services. Contextor can interface with both active and minimised programmes.

Nice Systems

The friendly RPA tool named NEVA-Nice Employee Virtual Attendant is an intelligent tool that assists in automating mundane tasks, compliance adherence, and Upsell. It provides cloud-based and on-premise solutions and attended and unattended server automation.

The trio, AI, ML and RPA, are separate entities, closely interconnected. As it can solve most real-world issues in a blink, they have become an inseparable helping hand in all the major businesses.

What is an SLM?

A Small Language Model (SLM) is tailored to excel in simpler tasks, offering boosted accessibility and user-friendliness for organizations operating with limited resources. Besides, they can be readily fine-tuned to align with specific requirements. Small language models are particularly well-suited for organizations aiming to develop applications capable of operating local devices instead of relying on cloud infrastructure. They are especially beneficial for tasks that do not necessitate extensive reasoning or immediate responses.

Reasons to use SLMs

Given the growing popularity and applicability of SLMs across various domains, particularly in areas like sustainability and the volume of data required for training, there are multiple reasons for employing them.

From a hardware perspective, utilizing SLMs is more cost-effective, requiring less computational power and memory. Moreover, their suitability for on-premises and on-device deployments enhances security measures.

Regarding usage, SLMs are compact language models, either trained or fine-tuned for specific domains or tasks. Consequently, they can incorporate specialized terminology and knowledge, from legal jargon to medical diagnoses, safeguarding intellectual property. Depending on the context, SLMs offer a more economical and efficient solution.

SLMs find applications across diverse sectors, encompassing healthcare, technology, and beyond. Their utility extends to text summarization, content generation, sentiment analysis, chatbot development, named entity recognition, spell checking, machine translation, and code generation.

What is Phi-3?

Microsoft has a suite of small language models (SLMs) known as 'Phi,' demonstrating outstanding performance across various benchmarks. Microsoft's recent release is Phi-3, a series of open AI models. The Phi-3 models represent a prototype of capability and cost-effectiveness among small language models (SLMs), exceeding models of equivalent and larger sizes across the spectrum of coding, language, reasoning, and mathematical standards. This launch broadens the array of high-calibre models accessible to customers, providing them with more practical options as they craft and construct generative AI applications.

Phi-3-mini, a 3.8B language model, is accessible through Microsoft Azure AI Studio, Hugging Face, and Ollama. It is offered in two context-length variations—4K and 128K tokens. Notably, it is the first model within its category to support a context window of up to 128K tokens with minimal impact on quality. Furthermore, it is instruction-tuned, implying that it has been trained to comprehend and adhere to diverse instructions, mirroring natural human communication patterns. This ensures that the model is readily deployable straight out of the box. Phi-3-mini is available on Azure AI to leverage the deploy-eval-finetune toolchain, and it is also accessible on Ollama for developers to execute locally on their laptops.

Features of Phi-3

Phi-3 models exhibit distinctive superiority over language models of comparable and larger dimensions on key benchmarks, showcasing the following features:

Suitability for resource-constrained environments, including on-device and offline inference scenarios.

Performance optimization for latency-bound scenarios where rapid response times are paramount.

Tailored for use cases with simpler tasks.

Utilization of high-quality training data to ensure model accuracy.

Enhancement through comprehensive safety measures post-training, including reinforcement learning from human feedback (RLHF), automated testing and evaluations across numerous harm categories, and manual red-teaming.

Adherence to the Microsoft Responsible AI Standard, embodying a company-wide adherence to six fundamental principles: accountability, transparency, fairness, reliability and safety, privacy and security, and inclusiveness.

Snowflake meets Phi-3: Advantages  

The key pain point about LLMs is the computing required to host and run them. Setting up a dozen GPUs to run models can be expensive and complex. There's where Snowflake steps up. Snowflake's compute pool option enables users to easily and quickly set up and manage compute clusters. Phi-3 comes into the picture because of its cost-effective GPU utilization.

Can you imagine a situation where your language model only requires less than 3GB of GPU memory for inference? Well, now it's possible, all thanks to Phi-3. It's a state-of-the-art SLM that produces excellent results over GP3.5 and Mistral 8x7B, which are much bigger models. This opens the door for more cost-effective solutions to be brought up in the AI space. Add Snowflake for hosting; you have an excellent setup to host, test, and build AI applications. Read below how Beinex managed to run Phi-3 on Day 0 in Snowflake.  Figure 1: DocAI running on Phi-3

Implementing Phi-3 on Snowflake: What Beinex Did and How Beinex Did it?

Beinex has seamlessly integrated Phi-3 into Snowflake to help enterprises unlock their data's full potential through advanced language processing capabilities and enhance decision-making with deeper insights. The integration facilitates Snowflake users to:

Capitalize on the efficiency of Phi-3 in environments with limited resources.

Ensure swift responses to queries and data requests, improving operational efficiency.

Achieve cost savings without compromising on model performance, optimizing resource allocation.

Boost the precision and dependability of their data-driven insights and predictions.

Leverage state-of-the-art AI capabilities while strictly adhering to data governance and compliance standards.

Here's a detailed guide on implementing Phi-3 on Snowflake:
Step 1: Create Necessary Objects  -- Run by ACCOUNTADMIN to allow connecting to Hugging Face to download the model  -- Stage to store LLM models  CREATE STAGE <stagename> IF NOT EXISTS models  DIRECTORY = (ENABLE = TRUE)  ENCRYPTION = (TYPE='SNOWFLAKE_SSE');    -- Stage to store YAML specs  CREATE STAGE <stagename> IF NOT EXISTS specs  DIRECTORY = (ENABLE = TRUE)  ENCRYPTION = (TYPE='SNOWFLAKE_SSE');    -- Image repository  CREATE OR REPLACE IMAGE REPOSITORY images;    -- Compute pool to run containers  CREATE COMPUTE POOL GPU_NV_S    MIN_NODES = 1    MAX_NODES = 1    INSTANCE_FAMILY = GPU_NV_S;  Step 2: Docker Image Code - ollama  FROM ollama/ollama    RUN $(ollama serve > output.log 2>&1 &) && sleep 10 && ollama pull phi3 && pkill ollama && rm output.log    ENTRYPOINT ["ollama"]  CMD ["serve"]  Step 3: Tag and Push the Docker Image  docker tag ollama <SNOW_ORG-SNOW_ACCOUNT>.registry.snowflakecomputing.com/db/schema/image respository /ollama  docker push <SNOW_ORG-SNOW_ACCOUNT>.registry.snowflakecomputing.com db/schema/image repository /ollama  Step 4: Docker Image - UDF  FROM python:3.11    WORKDIR /app  ADD ./requirements.txt /app/    RUN pip install --no-cache-dir -r requirements.txt    ADD ./ /app    EXPOSE 5000    ENV FLASK_APP=app  CMD ["flask", "run", "--host=0.0.0.0"]  App.py content is given below :  from flask import Flask, request, Response, jsonify  import logging  import re  import os  from openai import OpenAI    client = OpenAI(      base_url='http://ollama:11434/v1',      api_key="EMPTY",  )    model = "phi3"    app = Flask(__name__)  app.logger.setLevel(logging.ERROR)      def extract_json_from_string(s):      logging.info(f"Extracting JSON from string: {s}")      # Use a regular expression to find a JSON-like string      matches = re.findall(r"\{[^{}]*\}", s)        if matches:          # Return the first match (assuming there's only one JSON object embedded)          return matches[0]        # Return the original string if no JSON object is found      return s      @app.route("/", methods=["POST"])  def udf():      try:          request_data: dict = request.get_json(force=True)  # type: ignore          return_data = []            for index, col1 in request_data["data"]:              completion = client.chat.completions.create(                  model=model,                  messages=[                      {                          "role": "system",                          "content": "You are a bot to help extract data and should give professional responses",                      },                      {"role": "user", "content": col1},                  ],              )              return_data.append(                  [index, extract_json_from_string(completion.choices[0].message.content)]              )            return jsonify({"data": return_data})      except Exception as e:          app.logger.exception(e)          return jsonify(str(e)), 500  Step 6: YAML File  spec:    containers:      - name: ollama        image: <SNOW_ORG-SNOW_ACCOUNT>.registry.snowflakecomputing.com/ db/schema/image respository /Phi3        resources:          requests:            nvidia.com/gpu: 1          limits:            nvidia.com/gpu: 1        env:          NUM_GPU: 1          MAX_GPU_MEMORY: 24Gib        volumeMounts:          - name: llm-workspace            mountPath: /<stage name>      - name: udf        image: <SNOW_ORG-SNOW_ACCOUNT>.registry.snowflakecomputing.com/ db/schema/image respository /ollama_udf    endpoints:      - name: chat        port: 5000        public: false      - name: llm        port: 11434        public: false    volumes:      - name: llm-workspace        source:  "@<llm stage_name>"  Step 7: Upload YAML File and Create Service  Upload the YAML file to the created stage, where the stage name in the YAML file should match the stage created in Step 2.  -- Create service  create service phi3  IN COMPUTE POOL <name of compute pool created>  FROM @dash_stage  SPECIFICATION_FILE = '<name of yaml file uploaded>';  Step 8: Create Service Function  Create a service function on the service (after it starts).  create or replace function phi3chat(prompt text)  returns text  service= phi3  endpoint=chat;  Check Service Status  Use the following command to check the status of the service:  SELECT     v.value:containerName::varchar container_name,    v.value:status::varchar status,    v.value:message::varchar message  FROM (    SELECT parse_json(system$get_service_status('<service name>'))  ) t,   LATERAL FLATTEN(input => t.$1) v; 

Benefits of Running Phi-3 on Snowflake

1. Cost-Effectiveness and Efficiency:

Phi-3's efficacy as a model, coupled with Snowflake's cloud-based infrastructure, ensures cost-effective operations.

Running Phi-3 on Snowflake minimizes operational expenses by optimizing resources and maximizing performance while delivering exceptional results.

2. Compatibility with Smaller GPUs:

Phi-3's versatility allows it to run efficiently on smaller GPUs, making it accessible to a wider range of users.

This flexibility enables organizations to leverage existing hardware infrastructure without needing extensive upgrades.

3. Exceptional Performance:

Despite being an SLM, Phi-3 outperforms both SLMs and Large Language Models (LLMs) available in the market.

This superior performance translates to more accurate and reliable user outcomes, enhancing productivity and decision-making.

4. Faster Response Times:

Even in resource-constrained environments, Phi-3's capability to deliver faster responses is particularly advantageous within Snowflake's ecosystem.

This integration ensures swift data processing and analysis by minimizing latency and maximizing throughput, enabling users to make informed real-time decisions.

SLM vs LLM

The choice between small and large language models hinges on organizational needs, task complexity, and resource availability.

LLMs excel in applications requiring the orchestration of intricate tasks, encompassing advanced reasoning, data analysis, and contextual comprehension.

On the other hand, SLMs present viable options for regulated industries and sectors facing scenarios necessitating top-tier results while maintaining data within their premises.

Both large and small language models possess distinct strengths and applications. While large language models thrive in managing complex workflows, small language models deliver impressive performance despite their compact size.

While some customers may exclusively require small models, others may favour larger models, with many seeking to integrate both types in various configurations. Ultimately, the optimal selection depends on the unique context and objectives of the organization. Besides transitioning from large to small models, the trend is evolving towards a diversified portfolio of models. This means that instead of relying on a single model, customers can choose from various models with different sizes, capabilities, and resource requirements. This empowers customers to decide the best model for their scenario, balancing performance and resource constraints.

Bringing automation to data management is one of Tableau’s big focuses for this release, to help users get insights faster. Some key features that help with that are:

• Virtual Connections
• Centralized Row-Level Security
• Edit Published Data Source
• Parameters in Tableau Prep

Virtual Connections – Complete data source oversight in one place Virtual Connections provide users with a single centralized connection to store credentials and other connection information for a data source. It would easily connect to data as opposed to that of embedding it in all workbooks and data sources that use it. Along with greatly reducing the maintenance load, this feature enables us to create tables and share access while securing service account credentials, establishing Data Policies, and extracting data centrally. Virtual connections might sound like a published data source but in published data sources, database connection information (like credentials) is bundled with higher-level data modeling like joins, calculated fields, default properties, etc. So, any user who wants to create and share data models is also responsible for storing and sharing the connection information, even if they aren’t the best person to do so. Virtual Connections separate these two concerns into different objects so they can be owned and managed separately. Centralized Row-Level Security - Bring precision and agility to data protection In general, Row-Level Security (RLS) refers to filtering out rows of data at query time based on the current user's identity. Due to this, different users can view the same table, viz, or report and see only data they are authorized to see. RLS allows scenarios like having a regional manager only see sales created within their team, but not sales from another manager's team. Rather than implementing RLS individually on every Tableau workbook or data source accessing a sensitive table, virtual connections can be used to centrally define and manage data policies. These policies will be consistently applied across all connected Tableau Flows, Data Sources, or Workbooks that depend on that data. This gives granular control over data security while bringing flexibility to reuse data sources. Edit Published Data Source This feature makes managing data sources very effortless. Now we can create, edit and rename data sources directly in Tableau Server and Tableau Online, test the changes, and publish—all without leaving the browser. This streamlines the governance and metadata management. Parameters in Tableau Prep Parameters in Tableau Prep speed up and simplify tasks by enabling users to easily change data inputs, data outputs, and values used throughout a flow which makes reusing them much easier. Supported step types in 2021.4:

• Inputs /Outputs File path/File name
• Published data source name
• Calculated fields
• Filters

Copy and paste in dashboards Another feature that needs a special mention is Copy and paste in dashboards. Now users can easily copy images, text boxes, and web page containers within the same and across different dashboards and workbooks in Tableau Online, Tableau Server, and Tableau Desktop. Through this, one could develop customized and interactive dashboards rapidly by recreating the already formatted items. Tableau 2021.4 Highlights

1. Data Management
   1. Virtual Connections
   2. Centralized Row-Level Security
2. Tableau Catalog
   1. Inherited column descriptions
   2. Create flows from Catalog pages
3. Tableau Prep Builder
   1. Tableau Prep Parameters
   2. Create Tiles
   3. Flow output Subscriptions
4. Tableau Online
   1. Connected Apps
   2. Tableau Bridge Multi-Pool
5. Analytics
   1. Metric Updates
      1. Show Metric status with colors
      2. Customize comparison period and data window
      3. Embed metrics
   2. Replay Animations
   3. Web authoring improvements
      1. Edit published data sources on Server and Online
      2. Copy/Paste Dashboard Zones
   The journey towards unveiling the full potential of data analytics can be experienced in a better way with this new adoption of Tableau. Guaranteed to make your work polished and effortless, Tableau 2021.4 opens up its myriad capabilities to help you derive meaning out of data. Let all the tech geeks armor up to savour the essence of this exciting update.