DEV Community: JJ Dev

Run DeepSeek R1 with Ollama on Windows - 100% Local Solution

JJ Dev — Fri, 31 Jan 2025 08:24:56 +0000

On January 20, 2025, DeepSeek released its first reasoning model, DeepSeek-R1, which demonstrates performance comparable to OpenAI’s o1 model in mathematics, coding, and reasoning tasks.

Its open-source nature allows developers to freely use, modify, and commercialize it, which has garnered significant attention within the AI community. Additionally, the model achieves high efficiency at a lower cost and also provides distillation technology, enabling the transfer of knowledge from larger models to smaller ones, thereby reducing computational resource requirements.

DeepSeek-R1 Local Deployment

1. Ollama

Download the Ollama software: Download Ollama, double-click to install.

You can use the following command to check if Ollama is installed successfully:

ollama --version

2. Run the model

You can check the official website for the model and find the commands

A model with 1.5B (1.5 billion) parameters is considered a relatively smaller-scale LLM compared to larger models like 7B, 13B, or 65B parameters, which require significantly more computational resources.

Using a 1.5B parameter model is often a trade-off between efficiency and performance, making it more suitable for environments with limited hardware while still maintaining reasonable accuracy and capabilities.

ollama run deepseek-r1:1.5b

Now, we can try sending a message to the DeepSeek-local model.

>>> give me code about Python Program to Add Two Numbers

3. Result

When using DeepSeek-R1, the model generates not only answers but also intermediate steps that outline its logical reasoning.

<think>
Okay, the user asked for a Python program to add two numbers. I need to provide a clear and simple solution.

First, I'll start by defining two variables, a and b. Then, I'll prompt the user to input their values using the
input() function.

I should consider using float() to handle decimal inputs because some people might enter numbers with decimals.
That way, the addition is precise.

Next, I'll write the sum by adding a and b together and store it in a variable called total.

After calculating the sum, I'll display both original numbers and their sum. Using print() will make this easy for
the user to see.

I should also add an example so the user can test their own numbers. That way, they can see how the program works
when they run it.
</think>

Here's a simple Python program that adds two numbers:

# Define the two numbers
a = float(input("Enter first number: "))
b = float(input("Enter second number: "))

# Calculate the sum
total = a + b

# Display the result
print(f"The sum of {a} and {b} is {total}")

Explanation:

Define the numbers: The program first reads two numbers from the user using input() and converts them to floating-point numbers.
Calculate the sum: It adds the two numbers together and stores the result in a variable called total.
Display the result: Finally, it prints out the sum with a formatted string that includes both original numbers.

You can run this program by copying the code into a Python file and executing it using a text editor or the Python
command line.

4. Evaluation

The code generation is correct, with a total response time of 26 seconds.

CPU

Model: AMD Ryzen 7 3700X 8-Core Processor
Cores: 8 cores / 16 threads
Base Clock: 3.59 GHz

Memory

Capacity: 16GB

5. Stop the model

# List running models
ollama ps

# Stop a running model
ollama stop deepseek-r1:1.5b

❤️‍🔥 Thank you for reading my article. 🎉🎉

How do I troubleshoot issues about Amazon MSK cluster?

JJ Dev — Wed, 11 Sep 2024 15:00:35 +0000

1. Create Topic use SASL/SCRAM

When I ran this command to create a topic. These are some errors below.

bin/kafka-topics.sh --create --bootstrap-server <endpoint> --topic <topic_name> --command-config client.properties

[java.lang.OutOfMemoryError Java heap space]
[ERROR org.apache.kafka.common.errors.TimeoutException: The AdminClient thread has exited. Call: createTopics]

Soulation：
The error OutOfMemoryError occurred, which is a common issue when connecting to Kafka using SASL_SSL. This is not actually related to memory, but rather because the client was not configured to connect using SSL.
You should configure your admin client with --command-config <ssl.conf>

2. Public Access by SASL/SCRAM

If you are using the SASL/SCRAM or mTLS access-control methods, and want to enable public access msk. You will show the error.

Soulation：
After you set the Apache Kafka ACLs for your cluster, update the cluster's configuration.
Add Cluster configuration allow.everyone.if.no.acl.found=false

3. Consumers ACL rules

When you config allow.everyone.if.no.acl.found=false, must be set for specific consumer groups, and --group '*' cannot be used.

[org.apache.kafka.common.errors.GroupAuthorizationException: Not authorized to access group: console-consumer-1736]

Soulation：
Now, explicitly specify that user:user_name can read the topic:demo-topic and use group:test.

bin/kafka-acls.sh --command-config client_sasl.properties --bootstrap-server <endpoint> --add --allow-principal "User:user_name" --operation Read --group=test --topic demo-topic

Use EC2 user data script run a flask app🌐

JJ Dev — Mon, 19 Aug 2024 02:16:59 +0000

Creating Employee Directory Application automatically with AWS

🔥You can create an EC2 Amazon Linux 2023 user data script
It will run script after you start EC2.

Launch EC2 instance

select the Amazon Machine Image
Amazon Machine Images can come from AWS, like this Amazon Linux 2023 AMI, for example. This is the one that we will be using, but you can also browse more AMIs that would include AMIs coming from the AWS Marketplace.
just a lab, you can choose free tier eligible
Next, we have to decide if we want to configure a key pair.
We are going to select that we are going to proceed without a key pair for this, and the reason for that is because I don't intend to SSH into this instance to connect to it.
Network settings
Edit Keep the default VPC selection
Subnet: Choose the first subnet in the dropdown list
Auto-assign Public IP: Enable🖱️
Then expanding the advanced details section
under IAM instance profile choose S3DynamoDBFullAccessRole
pasting user data below (change AWS_DEFAULT_REGION)
click Launch instances

For Amazon Linux 2023

#!/bin/bash -ex
wget https://aws-tc-largeobjects.s3-us-west-2.amazonaws.com/DEV-AWS-MO-GCNv2/FlaskApp.zip
unzip FlaskApp.zip
cd FlaskApp/
yum -y install python3-pip
pip install -r requirements.txt
yum -y install stress
export PHOTOS_BUCKET=${SUB_PHOTOS_BUCKET}
export AWS_DEFAULT_REGION=ap-northeast-1
export DYNAMO_MODE=on
FLASK_APP=application.py /usr/local/bin/flask run --host=0.0.0.0 --port=80

For Amazon Linux 2

#!/bin/bash -ex
wget https://aws-tc-largeobjects.s3-us-west-2.amazonaws.com/DEV-AWS-MO-GCNv2/FlaskApp.zip
unzip FlaskApp.zip
cd FlaskApp/
yum -y install python3 mysql
pip3 install -r requirements.txt
amazon-linux-extras install epel
yum -y install stress
export PHOTOS_BUCKET=${SUB_PHOTOS_BUCKET}
export AWS_DEFAULT_REGION=<INSERT REGION HERE>
export DYNAMO_MODE=on
FLASK_APP=application.py /usr/local/bin/flask run --host=0.0.0.0 --port=80

open address🔚

your Employee Directory Application running~
when you start the instance, the script will execute.

AWS Cloud Quest🎮: EFS mount on EC2

JJ Dev — Wed, 24 Jul 2024 16:44:04 +0000

Topic: File Systems in the Cloud

🚩issue: Web servers in different Availability Zones need to access the same file data.

Step 1. Create security groups for EFS✨

Create security groups for EFS → Allow web servers to access EFS

Security Group：Acts as a virtual firewall, used to control inbound and outbound traffic for EC2 instances. Can be used within an existing VPC.

In the lab, choose VPC of PetModelsWebServer.
Set the NFS type in the Inbound Rule of the Security Group of EFS
Choose the Security Group of PetModelsWebServer as source
Only resources belonging to the Web server Security Group are allowed to access EFS

By selecting a security group as the incoming source, any EC2 instances linked to the security group you select will have NFS client access to the file system.

Step 2. Create an EFS✨

Create File System on the EFS security group
Set network access to allow mount Target
- Az-1 to EFS security groups
- Az-2 to EFS security groups
- Az-3 to EFS security groups
click Attach button to copy mount command

Step 3. EC2 mount NFS✨

Ok! Now, all environments are ready.
You will mount a /data folder on EC2.
The following are the command for mounting a NFS in Linux

sudo -i

# download aws efs utils
sudo yum install -y Amazon-efs-utils

# create folder
mkdir data

# paste from "Attach" button
sudo mount -t efs -o tls fs-id:/ data

cd data

# write text in file
sudo bash -c “cat >> efs-1-setup.log”

# cat -> output "efs-1-setup.log"

efs-1 mounted in site A

cat efs-1-setup.log

Step 4. Repeat Step 3. for each EC2✨

After setting, you can read and write files from other EC2s.
All changes will be synchronized with EFS.