Pakistan's First Oracle Blog

 This video is a hands-on step-by-step primer about how to quantize any model using Hugging Face Quanto which is a versatile pytorch quantization toolkit.

!pip install transformers==4.35.0

!pip install quanto==0.0.11

!pip install torch==2.1.1

!pip install sentencepiece==0.2.0

model_name = "google/flan-t5-small"

import sentencepiece as spm

from transformers import T5Tokenizer, T5ForConditionalGeneration

tokenizer = T5Tokenizer.from_pretrained("google/flan-t5-small")

model = T5ForConditionalGeneration.from_pretrained("google/flan-t5-small")

input_text = "Meaning of happiness is "

input_ids = tokenizer(input_text, return_tensors="pt").input_ids

outputs = model.generate(input_ids)

print(tokenizer.decode(outputs[0]))

from helper import compute_module_sizes

module_sizes = compute_module_sizes(model)

print(f"The model size is {module_sizes[''] * 1e-9} GB")

from quanto import quantize, freeze

import torch

quantize(model, weights=torch.int8, activations=None)

freeze(model)

module_sizes = compute_module_sizes(model)

print(f"The model size is {module_sizes[''] * 1e-9} GB")

input_text = "Meaning of happiness is "

input_ids = tokenizer(input_text, return_tensors="pt").input_ids

outputs = model.generate(input_ids)

print(tokenizer.decode(outputs[0]))

 This video is a hands-on step-by-step primer about how to use RAG with Open AI File Search. OpenAI now supports RAG which means that now you can attach your own files and custom data to OpenAI assistant and talk to your documents with GPT4. 

Make sure you have installed latest version of openai on your local system:

pip install openai --upgrade

also make sure to have data.txt in the same folder as your script. 

 Function calling in AI simply means that you can call external APIs from within your AI-powered application. Whenever you read that a model can do function calling, it means that it can take a natural language query of user and convert it to a function call. Then you can execute that function call to your API endpoint to get the data, and give it to LLM as additional context and get more grounded latest response as per your application requirement.

This video shows how to locally install AnythingLLM to privately and securly and remotely run any LLM with any RAG document. It all runs locally with zero required internet connectivity.

 Artificial intelligence (AI) is revolutionizing our interactions with the world, presenting both opportunities and complexities for organizations embracing new AI technologies. Nowhere is this more evident than with generative AI, which integrates knowledge assimilation from various sources to automate tasks and boost human creativity and productivity. This transformative technology empowers organizations to summarize documents, create tables, generate meaningful text, develop code, and synthesize innovative ideas.

Oracle's strategic approach revolves around three AI modalities: infrastructure, models and services, and application integration. Oracle has embedded AI across its cloud applications like CRM, ERP, HCM, CX, and EMR to enhance their functionalities. Similarly, Oracle is introducing generative AI capabilities into its database portfolio, akin to its integration of machine learning (ML) features into Oracle Database service and MySQL HeatWave.

Oracle Cloud Infrastructure (OCI) Generative AI Agents harnesses the potential of large language models (LLMs) and retrieval-augmented generation (RAG), enabling users to query extensive enterprise knowledge bases. Soon, users will access real-time information via natural language interfaces for direct action.

OCI Generative AI Agents RAG service empowers employees to glean insights and answers from disparate sources within the company through natural conversations. The service matches semantic similarities between user queries and documents in a centralized embeddings vector store, consolidating data from diverse corporate systems like ERP, HCM, CX, and data lakes. Users receive results using the most current information, even from dynamic data stores, rather than static reports or outdated indexes.

At the heart of this solution is a large language models or LLMs in short, facilitating natural, conversational interactions with users. You can either use Meta's Llama model or a model from Cohere and that list is going to grow. Think of it as a knowledgeable colleague you can ask business questions in everyday language. This is complemented by an embeddings model that generates embeddings vectors for the corporate document corpus stored in the embeddings vector database. 

The retrieval-augmented generation (RAG) system, comprising the vector database, embeddings model, and text generation model, synthesizes responses grounded in retrieved documents. Instead of presenting a list of documents, the AI service delivers natural language responses, supported with references and links to original sources. In the future, users will seamlessly transition from information retrieval to immediate actions like sending emails, contacting customers, or managing orders within the same interface.

Select AI is probably one of the most magical AI feature in Oracle Autonomous database. I have been using it for some time now and it's feels simply like magic.

Select AI represents a cutting-edge feature within the Oracle Autonomous Database, introduced last year, that empowers users to interact with their data using natural language queries. This innovative capability enables individuals to pose questions and receive insights without needing to comprehend the underlying data storage or SQL formulation. By simply directing queries in languages like English, Spanish, French, Portuguese, and more, users can leverage the full potential of their database.

The underlying technology behind Select AI leverages advanced generative AI models, such as those developed by OpenAI or Cohere, which excel at understanding intent and generating SQL code based on natural language prompts. This approach allows developers to extend and build new natural language applications seamlessly within the SQL framework. Select AI streamlines this process by integrating with Oracle SQL, utilizing a standard SELECT statement followed by AI and the natural language query.

One of the remarkable features of Select AI is its conversational capabilities, enhanced with recent updates. Users can engage in a dialogue with the database, posing follow-up questions that build upon previous interactions. This chat history is leveraged by the AI models to provide context-aware responses, resembling a real-life conversation that facilitates deeper exploration of data insights.

Moreover, Select AI is designed to be flexible and adaptable to various business needs. It supports multiple Large Language Models (LLMs) including Cohere and OpenAI, with potential for expansion to include domain-specific models in the future. This pluggable design allows organizations to select the most suitable LLM for their specific use cases, ensuring optimal performance and alignment with business objectives.

With Autonomous Database Select AI, developers can create AI-enabled applications that comprehend users' natural language queries within a conversational context, enabling secure and efficient access to valuable insights from private data sources. This groundbreaking capability marks a significant leap forward in data analytics, harnessing the power of generative AI and natural language processing to democratize access to database insights.

This is a step by step easy tutorial to locally install OpenDevin which is an open-source project aiming to replicate Devin, an autonomous AI software engineer who is capable of executing complex engineering tasks and collaborating actively with users on software development projects.

Commands Used:

Pre-requisites:

Update OS:

sudo apt-get update

Install git:

sudo apt-get install git

Install Nodejs:

sudo apt-get install nodejs

Install npm:

sudo apt install npm

Install Rust:

curl --proto '=https' --tlsv1.3 https://sh.rustup.rs -sSf | sh

Install docker:

sudo apt-get update

sudo apt-get install ca-certificates curl

sudo install -m 0755 -d /etc/apt/keyrings

sudo curl -fsSL https://download.docker.com/linux/ubuntu/gpg -o /etc/apt/keyrings/docker.asc

sudo chmod a+r /etc/apt/keyrings/docker.asc

# Add the repository to Apt sources:

echo \

  "deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/docker.asc] https://download.docker.com/linux/ubuntu \

  $(. /etc/os-release && echo "$VERSION_CODENAME") stable" | \

  sudo tee /etc/apt/sources.list.d/docker.list > /dev/null

sudo apt-get update

sudo apt-get install docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin

Install conda:

wget https://repo.anaconda.com/archive/Anaconda3-2022.05-Linux-x86_64.sh

bash Anaconda3-2022.05-Linux-x86_64.sh

Install uvicorn:

sudo apt install uvicorn

Backend Commands:

git clone https://github.com/OpenDevin/OpenDevin.git

cd OpenDevin

conda create -n opendevin python=3.11

conda activate opendevin 

docker ps

docker pull ghcr.io/opendevin/sandbox

which python

python --version

/home/ubuntu/anaconda3/envs/opendevin/bin/python -m pip install pipenv

/home/ubuntu/anaconda3/envs/opendevin/bin/python -m pipenv install -v

/home/ubuntu/anaconda3/envs/opendevin/bin/python -m pipenv shell

uvicorn opendevin.server.listen:app --port 3000

Frontend Commands:

python -m pipenv requirements > requirements.txt && python -m pip install -r requirements.txt

PYTHONPATH=`pwd` /home/ubuntu/anaconda3/envs/opendevin/bin/python opendevin/main.py -d ./workspace/ -i 100 -t "Write a bash script that prints 'hello world'"

 The Oracle Resource Manager is a powerful tool for administrators to maintain system stability, optimize performance, and manage workload prioritization within Oracle Database environments. It provides granular control over resource allocation and ensures that database resources are utilized efficiently to meet business requirements and service level objectives.

Resource Manager is a feature that enables administrators to manage and prioritize system resources such as CPU and memory among different users, applications, sessions, and consumer groups within the database. The Resource Manager allows for the control of resource allocation based on specified policies, ensuring that critical workloads receive adequate resources while preventing resource contention and maintaining overall system performance.

This Oracle script is setting up a comprehensive resource management strategy using Oracle Resource Manager. It defines consumer groups to classify different types of workloads and establishes resource plans with directives to allocate resources and prioritize workload processing based on specified criteria like workload type, time of day, and user groups. This provided Oracle PL/SQL script demonstrates the configuration and utilization of Oracle Resource Manager to manage database resource allocation based on workload characteristics. 

PROCEDURE create_consumer_group(

  consumer_group  IN  VARCHAR2,

  comment         IN  VARCHAR2,

  cpu_mth         IN  VARCHAR2 DEFAULT 'ROUND-ROBIN')

The create_consumer_groups.sql script uses this procedure to create the OLTP and batch consumer groups.

CONN sys AS SYSDBA

BEGIN

  DBMS_RESOURCE_MANAGER.clear_pending_area;

  DBMS_RESOURCE_MANAGER.create_pending_area;

  -- Create the consumer groups

  DBMS_RESOURCE_MANAGER.create_consumer_group(

    consumer_group => 'oltp_consumer_group',

    comment        => 'OLTP process consumer group.');

  DBMS_RESOURCE_MANAGER.create_consumer_group(

    consumer_group => 'batch_consumer_group',

    comment        => 'Batch process consumer group.');

  DBMS_RESOURCE_MANAGER.validate_pending_area;

  DBMS_RESOURCE_MANAGER.submit_pending_area;

END;

/

BEGIN

  DBMS_RESOURCE_MANAGER.clear_pending_area();

  DBMS_RESOURCE_MANAGER.create_pending_area();

  -- Delete consumer groups.

  DBMS_RESOURCE_MANAGER.delete_consumer_group (

    consumer_group => 'oltp_consumer_group'); 

  DBMS_RESOURCE_MANAGER.delete_consumer_group (

    consumer_group => 'batch_consumer_group'); 

  DBMS_RESOURCE_MANAGER.validate_pending_area;

  DBMS_RESOURCE_MANAGER.submit_pending_area();

END;

/

PROCEDURE create_plan (

  plan                      IN  VARCHAR2,

  comment                   IN  VARCHAR2,

  cpu_mth                   IN  VARCHAR2 DEFAULT 'EMPHASIS',

  active_sess_pool_mth      IN  VARCHAR2 DEFAULT 'ACTIVE_SESS_POOL_ABSOLUTE',

  parallel_degree_limit_mth IN  VARCHAR2 DEFAULT 'PARALLEL_DEGREE_LIMIT_ABSOLUTE',

  queueing_mth              IN  VARCHAR2 DEFAULT 'FIFO_TIMEOUT')                       

PROCEDURE create_plan_directive (

  plan                      IN  VARCHAR2,

   group_or_subplan          IN  VARCHAR2,

  comment                   IN  VARCHAR2,

  cpu_p1                    IN  NUMBER DEFAULT NULL,

  cpu_p2                    IN  NUMBER DEFAULT NULL,

  cpu_p3                    IN  NUMBER DEFAULT NULL,

  cpu_p4                    IN  NUMBER DEFAULT NULL,

  cpu_p5                    IN  NUMBER DEFAULT NULL,

  cpu_p6                    IN  NUMBER DEFAULT NULL,

  cpu_p7                    IN  NUMBER DEFAULT NULL,

  cpu_p8                    IN  NUMBER DEFAULT NULL,

  active_sess_pool_p1       IN  NUMBER DEFAULT NULL,

  queueing_p1               IN  NUMBER DEFAULT NULL,

  parallel_degree_limit_p1  IN  NUMBER DEFAULT NULL,

   switch_group              IN  VARCHAR2 DEFAULT NULL,

  switch_time               IN  NUMBER DEFAULT NULL,

  switch_estimate           IN  BOOLEAN DEFAULT FALSE,

  max_est_exec_time         IN  NUMBER DEFAULT NULL,

  undo_pool                 IN  NUMBER DEFAULT NULL,

  max_idle_time             IN  NUMBER DEFAULT NULL,

  max_idle_blocker_time     IN  NUMBER DEFAULT NULL,

  switch_time_in_call       IN  NUMBER DEFAULT NULL)

  BEGIN

  DBMS_RESOURCE_MANAGER.clear_pending_area;

  DBMS_RESOURCE_MANAGER.create_pending_area;

  -- Create a new plan

  DBMS_RESOURCE_MANAGER.create_plan(

    plan    => 'day_plan',

    comment => 'Plan suitable for daytime processing.');

  -- Assign consumer groups to plan and define priorities

  DBMS_RESOURCE_MANAGER.create_plan_directive (

    plan             => 'day_plan',

    group_or_subplan => 'oltp_consumer_group',

     comment          => 'Give OLTP processes higher priority - level 1',

    cpu_p1           => 80,

    switch_group     => 'batch_consumer_group',

    switch_time      => 60);

  DBMS_RESOURCE_MANAGER.create_plan_directive (

    plan             => 'day_plan',

    group_or_subplan => 'batch_consumer_group',

    comment          => 'Give batch processes lower priority - level 2',

    cpu_p2           => 100);

  DBMS_RESOURCE_MANAGER.create_plan_directive(

    plan             => 'day_plan',

    group_or_subplan => 'OTHER_GROUPS',

    comment          => 'all other users - level 3',

    cpu_p3           => 100);

  DBMS_RESOURCE_MANAGER.validate_pending_area;

  DBMS_RESOURCE_MANAGER.submit_pending_area;

END;

/

BEGIN

  DBMS_RESOURCE_MANAGER.clear_pending_area;

  DBMS_RESOURCE_MANAGER.create_pending_area;

  -- Create a new plan

  DBMS_RESOURCE_MANAGER.create_plan(

    plan    => 'night_plan',

    comment => 'Plan suitable for daytime processing.');

  -- Assign consumer groups to plan and define priorities

  DBMS_RESOURCE_MANAGER.create_plan_directive (

    plan             => 'night_plan',

    group_or_subplan => 'batch_consumer_group',

    comment          => 'Give batch processes lower priority - level 2',

    cpu_p1           => 80);

  DBMS_RESOURCE_MANAGER.create_plan_directive (

    plan             => 'night_plan',

    group_or_subplan => 'oltp_consumer_group',

    comment          => 'Give OLTP processes higher priority - level 1',

    cpu_p2           => 100);

  DBMS_RESOURCE_MANAGER.create_plan_directive(

    plan             => 'night_plan',

    group_or_subplan => 'OTHER_GROUPS',

    comment          => 'all other users - level 3',

    cpu_p3           => 100);

  DBMS_RESOURCE_MANAGER.validate_pending_area;

  DBMS_RESOURCE_MANAGER.submit_pending_area;

END;

/

BEGIN

DBMS_RESOURCE_MANAGER.CREATE_PENDING_AREA();

DBMS_RESOURCE_MANAGER.CREATE_PLAN(PLAN => 'bugdb_plan', 

   COMMENT => 'Resource plan/method for bug users sessions');

DBMS_RESOURCE_MANAGER.CREATE_PLAN(PLAN => 'maildb_plan', 

   COMMENT => 'Resource plan/method for mail users sessions');

DBMS_RESOURCE_MANAGER.CREATE_PLAN(PLAN => 'mydb_plan', 

   COMMENT => 'Resource plan/method for bug and mail users sessions');

DBMS_RESOURCE_MANAGER.CREATE_CONSUMER_GROUP(CONSUMER_GROUP => 'Online_group', 

   COMMENT => 'Resource consumer group/method for online bug users sessions');

DBMS_RESOURCE_MANAGER.CREATE_CONSUMER_GROUP(CONSUMER_GROUP => 'Batch_group', 

   COMMENT => 'Resource consumer group/method for batch job bug users sessions');

DBMS_RESOURCE_MANAGER.CREATE_CONSUMER_GROUP(CONSUMER_GROUP => 'Bug_Maint_group',

   COMMENT => 'Resource consumer group/method for users sessions for bug db maint');

DBMS_RESOURCE_MANAGER.CREATE_CONSUMER_GROUP(CONSUMER_GROUP => 'Users_group', 

   COMMENT => 'Resource consumer group/method for mail users sessions');

DBMS_RESOURCE_MANAGER.CREATE_CONSUMER_GROUP(CONSUMER_GROUP => 'Postman_group',

   COMMENT => 'Resource consumer group/method for mail postman');

DBMS_RESOURCE_MANAGER.CREATE_CONSUMER_GROUP(CONSUMER_GROUP => 'Mail_Maint_group', 

   COMMENT => 'Resource consumer group/method for users sessions for mail db maint');

DBMS_RESOURCE_MANAGER.CREATE_PLAN_DIRECTIVE(PLAN => 'bugdb_plan',

   GROUP_OR_SUBPLAN => 'Online_group',

   COMMENT => 'online bug users sessions at level 1', MGMT_P1 => 80, MGMT_P2=> 0);

DBMS_RESOURCE_MANAGER.CREATE_PLAN_DIRECTIVE(PLAN => 'bugdb_plan', 

   GROUP_OR_SUBPLAN => 'Batch_group', 

   COMMENT => 'batch bug users sessions at level 1', MGMT_P1 => 20, MGMT_P2 => 0,

   PARALLEL_DEGREE_LIMIT_P1 => 8);

DBMS_RESOURCE_MANAGER.CREATE_PLAN_DIRECTIVE(PLAN => 'bugdb_plan', 

   GROUP_OR_SUBPLAN => 'Bug_Maint_group',

   COMMENT => 'bug maintenance users sessions at level 2', MGMT_P1 => 0, MGMT_P2 => 100);

DBMS_RESOURCE_MANAGER.CREATE_PLAN_DIRECTIVE(PLAN => 'bugdb_plan', 

   GROUP_OR_SUBPLAN => 'OTHER_GROUPS', 

   COMMENT => 'all other users sessions at level 3', MGMT_P1 => 0, MGMT_P2 => 0,

   MGMT_P3 => 100);

DBMS_RESOURCE_MANAGER.CREATE_PLAN_DIRECTIVE(PLAN => 'maildb_plan', 

   GROUP_OR_SUBPLAN => 'Postman_group',

   COMMENT => 'mail postman at level 1', MGMT_P1 => 40, MGMT_P2 => 0);

DBMS_RESOURCE_MANAGER.CREATE_PLAN_DIRECTIVE(PLAN => 'maildb_plan',

   GROUP_OR_SUBPLAN => 'Users_group',

   COMMENT => 'mail users sessions at level 2', MGMT_P1 => 0, MGMT_P2 => 80);

DBMS_RESOURCE_MANAGER.CREATE_PLAN_DIRECTIVE(PLAN => 'maildb_plan',

   GROUP_OR_SUBPLAN => 'Mail_Maint_group',

   COMMENT => 'mail maintenance users sessions at level 2', MGMT_P1 => 0, MGMT_P2 => 20);

DBMS_RESOURCE_MANAGER.CREATE_PLAN_DIRECTIVE(PLAN => 'maildb_plan',

   GROUP_OR_SUBPLAN => 'OTHER_GROUPS', 

   COMMENT => 'all other users sessions at level 3', MGMT_P1 => 0, MGMT_P2 => 0,

   MGMT_P3 => 100);

DBMS_RESOURCE_MANAGER.CREATE_PLAN_DIRECTIVE(PLAN => 'mydb_plan', 

   GROUP_OR_SUBPLAN => 'maildb_plan', 

   COMMENT=> 'all mail users sessions at level 1', MGMT_P1 => 30);

DBMS_RESOURCE_MANAGER.CREATE_PLAN_DIRECTIVE(PLAN => 'mydb_plan', 

   GROUP_OR_SUBPLAN => 'bugdb_plan', 

   COMMENT => 'all bug users sessions at level 1', MGMT_P1 => 70);

DBMS_RESOURCE_MANAGER.VALIDATE_PENDING_AREA();

DBMS_RESOURCE_MANAGER.SUBMIT_PENDING_AREA();

END;

/

 When managing Oracle temporary tablespaces, it's important to prioritize timely data commitment or rollback to release temporary space efficiently. 

Additionally, keep an eye out for sessions that remain active without progressing for an extended period; terminating these sessions may be necessary to optimize resource usage. Another crucial consideration is the allocation of small sport segments at the end of a temporary file, which can hinder successful file shrinking. To gain insights into this issue, check the High Water Mark (HWM) of the file to understand its allocation status and potential space usage implications. These practices help maintain the performance and manageability of Oracle temporary tablespaces effectively.

Following are some queries to manage these Temp tablespaces efficiently:

Use this query to get Temp tablespace usage:

SQL> select FILE#, STATUS,  Bytes/1024/1024 Size, tf.name Filename, ts.name TS_NAME  

    from v$tempfile tf , v$tablespace ts where tf.ts# = ts.ts#;

Use this query to get allocated temp space:

SQL> SELECT a.username, a.sid, a.serial#, a.osuser, b.tablespace, b.blocks, c.sql_text

      FROM v$session a, v$tempseg_usage b, v$sqlarea

  WHERE a.saddr = b.session_addr AND c.address= a.sql_address

      AND c.hash_value = a.sql_hash_value ORDER BY b.tablespace, b.blocks;

  

Use this query to get high water mark of Temp Tablespaces:

SQL>select sum( u.blocks * blk.block_size)/1024/1024 "MB. in sort segments", (hwm.max * blk.block_size)/1024/1024 "HWM"

    from v$sort_usage u, (select block_size from dba_tablespaces

     where contents = 'TEMPORARY') blk, 

(select segblk#+blocks max from

     v$sort_usage where segblk# = (select max(segblk#)

from v$sort_usage) ) hwm 

group by hwm.max * blk.block_size/1024/1024;   

 

Use this query to shrink the Temp Tablespaces:

SQL> alter tablespace temp shrink space;  

 This video is hands on step-by-step tutorial to create a new dataset, an AI model, fine-tune the model on dataset and then push it to hugging face.

Code:

%%capture

import torch

major_version, minor_version = torch.cuda.get_device_capability()

# Must install separately since Colab has torch 2.2.1, which breaks packages

!pip install "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"

if major_version >= 8:

    # Use this for new GPUs like Ampere, Hopper GPUs (RTX 30xx, RTX 40xx, A100, H100, L40)

    !pip install --no-deps packaging ninja flash-attn xformers trl peft accelerate bitsandbytes

else:

    # Use this for older GPUs (V100, Tesla T4, RTX 20xx)

    !pip install --no-deps xformers trl peft accelerate bitsandbytes

pass

!pip install einops

from unsloth import FastLanguageModel

import torch

max_seq_length = 2048 # Choose any! We auto support RoPE Scaling internally!

dtype = None # None for auto detection. Float16 for Tesla T4, V100, Bfloat16 for Ampere+

load_in_4bit = True # Use 4bit quantization to reduce memory usage. Can be False.

model, tokenizer = FastLanguageModel.from_pretrained(

    model_name = "unsloth/gemma-7b-bnb-4bit", # Choose ANY! eg teknium/OpenHermes-2.5-Mistral-7B

    max_seq_length = max_seq_length,

    dtype = dtype,

    load_in_4bit = load_in_4bit,

    token = " ", # use one if using gated models like meta-llama/Llama-2-7b-hf

)

model = FastLanguageModel.get_peft_model(

    model,

    r = 16, # Choose any number > 0 ! Suggested 8, 16, 32, 64, 128

    target_modules = ["q_proj", "k_proj", "v_proj", "o_proj",

                      "gate_proj", "up_proj", "down_proj",],

    lora_alpha = 16,

    lora_dropout = 0, # Supports any, but = 0 is optimized

    bias = "none",    # Supports any, but = "none" is optimized

    use_gradient_checkpointing = True,

    random_state = 3407,

    use_rslora = False,  # We support rank stabilized LoRA

    loftq_config = None, # And LoftQ

)

alpaca_prompt = """ذیل میں ایک ہدایت ہے جو فلم کے نام کی وضاحت کرتی ہے، اس کے ساتھ ایک ان پٹ بھی ہے جو مزید دستاویزات فراہم کرتا ہے۔ گانے کے بول لکھنے کے لیے ایک لمحہ نکالیں جو فلم کے نام کے معنی سے میل کھاتا ہے۔

### Instruction:

{}

### Input:

{}

### Response:

{}"""

EOS_TOKEN = tokenizer.eos_token # Must add EOS_TOKEN

def formatting_prompts_func(examples):

    instructions = examples["urdu_instruction"]

    inputs       = examples["urdu_input"]

    outputs      = examples["urdu_output"]

    texts = []

    for instruction, input, output in zip(instructions, inputs, outputs):

        # Must add EOS_TOKEN, otherwise your generation will go on forever!

        text = alpaca_prompt.format(instruction, input, output) + EOS_TOKEN

        texts.append(text)

    return { "text" : texts, }

pass

from datasets import load_dataset

dataset = load_dataset("fahdmirzac/urdu_bollywood_songs_dataset", split = "train")

dataset = dataset.map(formatting_prompts_func, batched = True,)

from huggingface_hub import login

access_token = "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"

login(token=access_token)

from trl import SFTTrainer

from transformers import TrainingArguments

trainer = SFTTrainer(

    model = model,

    tokenizer = tokenizer,

    train_dataset = dataset,

    dataset_text_field = "text",

    max_seq_length = max_seq_length,

    dataset_num_proc = 2,

    packing = False, # Can make training 5x faster for short sequences.

    args = TrainingArguments(

        per_device_train_batch_size = 2,

        gradient_accumulation_steps = 4,

        warmup_steps = 5,

        max_steps = 100,

        learning_rate = 2e-4,

        fp16 = not torch.cuda.is_bf16_supported(),

        bf16 = torch.cuda.is_bf16_supported(),

        logging_steps = 1,

        optim = "adamw_8bit",

        weight_decay = 0.01,

        lr_scheduler_type = "linear",

        seed = 3407,

        output_dir = "outputs",

    ),

)

trainer_stats = trainer.train()

FastLanguageModel.for_inference(model) # Enable native 2x faster inference

inputs = tokenizer(

[

    alpaca_prompt.format(

        "دیے گئے فلم کے نام کے بارے میں ایک مختصر گیت کے بول لکھیں۔", # instruction

        "کیوں پیار ہو گیا", # input

        "", # output - leave this blank for generation!

    )

], return_tensors = "pt").to("cuda")

outputs = model.generate(**inputs, max_new_tokens = 200, use_cache = True)

tokenizer.batch_decode(outputs)

FastLanguageModel.for_inference(model) # Enable native 2x faster inference

inputs = tokenizer(

[

    alpaca_prompt.format(

        "دیے گئے فلم کے نام کے بارے میں ایک مختصر گیت کے بول لکھیں۔", # instruction

        "رنگ", # input

        "", # output - leave this blank for generation!

    )

], return_tensors = "pt").to("cuda")

outputs = model.generate(**inputs, max_new_tokens = 200, use_cache = True)

tokenizer.batch_decode(outputs)

model.push_to_hub("fahdmirzac/Gemma_Urdu_Hindi_Bollywood_Songs", token = "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")

 This video is a hands-on guide as how to use vision features of Anthropic's Claude 3 Haiku AI model with Amazon Bedrock.

Code Used:

 This video is a step-by-step tutorial with code as how to create Amazon Bedrock AI agents with boto3 in Python to integrate with Lambda.

Code used: Just use any lambda with it of your choice.

 Blockchain tables in Oracle Database implement a unique data organization model, forming chains of insert-only rows. Each row, excluding the first, links back to the previous row using cryptographic hashing. This hashing method leverages row data and the hash of the preceding row to identify any unauthorized modifications through specialized verification procedures. The added capability to store user signatures significantly enhances security against tampering and unauthorized alterations.

Beyond thwarting unauthorized changes initiated by rogue or compromised insiders, blockchain tables boast powerful functionality:

• Identification of illicit modifications made by circumventing Oracle Database protocols.
• Detection of user impersonation and unauthorized data insertions on behalf of another user.
• Safeguarding against data tampering, certifying that all data inserted into the table was done legitimately.
• The chaining mechanism ensures that each row remains linked to its predecessor through cryptographic hashing. Altering any row disrupts this chain, immediately signaling tampering. However, this additional layer of security necessitates increased processing time during commit operations to chain rows effectively.

Let's explore how to create and interact with a blockchain table in Oracle Database through the following example:

-- Creating a blockchain table

SQL> CREATE BLOCKCHAIN TABLE secure_data_table (data_value VARCHAR2(128), sequence_number NUMBER)

     NO DROP UNTIL 1 DAYS IDLE

     NO DELETE UNTIL 16 DAYS AFTER INSERT

     HASHING USING "SHA2_512" VERSION "v1";    

Table created.

-- Inserting data into the blockchain table

SQL> INSERT INTO secure_data_table VALUES ('Value1', 1);

1 row created.

SQL> COMMIT;

Commit complete.

-- Querying metadata of the blockchain table

SQL> SELECT column_id, column_name, nullable, hidden_column

     FROM all_tab_cols

     WHERE table_name LIKE 'SECURE_DATA_%'

     ORDER BY column_id;

-- Viewing blockchain table details

SQL> SELECT * FROM dba_blockchain_tables;

In this example, secure_data_table is our blockchain table, ensuring immutable data integrity and providing robust security against unauthorized changes. The SHA2_512 hashing algorithm is employed to maintain the integrity of each chained row, guaranteeing data authenticity and traceability within Oracle Database's secure environment.

 If you want to access any object stored in an AWS S3 bucket from Oracle database, then dbms_cloud package is your friend, provided you are using Oracle Autonomous database. 

An AWS S3 (Simple Storage Service) bucket is a scalable storage resource in the cloud provided by Amazon Web Services. It allows you to store and retrieve any amount of data from anywhere on the web. S3 buckets are highly durable, secure, and can be configured to support various use cases including website hosting, data backup, and content distribution.

The DBMS_CLOUD package in Oracle Database is a PL/SQL package that provides procedures to interact with Oracle Cloud Infrastructure services, specifically for loading data into Oracle Database from external data sources such as object storage in Oracle Cloud. It enables seamless integration between Oracle Database and cloud-based storage for data loading operations, making it easier to ingest large volumes of data into the database from cloud-based sources. The package includes procedures like COPY_DATA for copying data from cloud storage to database tables, and CREATE_CREDENTIAL for managing credentials needed to access cloud storage securely. This package simplifies data loading tasks and enhances integration capabilities between Oracle Database and Oracle Cloud services.

In order to access the object from S3 bucket, make sure you have created a database user in Oracle with proper credentials, and also you have AWS IAM credentials to access S3 bucket.

Step 1: First, you need to create a credential in Oracle Database that allows access to your AWS S3 bucket. This credential will store the AWS access key ID and secret access key securely.

BEGIN

  DBMS_CLOUD.CREATE_CREDENTIAL(

    credential_name => 'MY_AWS_CREDENTIAL',

    username => 'AWS_ACCESS_KEY_ID',

    password => 'AWS_SECRET_ACCESS_KEY'

  );

END;

/

Step 2: Now, you can use the DBMS_CLOUD package to select data from a file in your AWS S3 bucket and load it into a table in Oracle Database.

DECLARE

  v_data CLOB;

BEGIN

  v_data := DBMS_CLOUD.GET_OBJECT(

    credential_name => 'MY_AWS_CREDENTIAL',

    object_uri => 'https://s3.<region>.amazonaws.com/<bucket_name>/<file_key>'

  );

  -- Process the retrieved data (example: print the data)

  DBMS_OUTPUT.PUT_LINE(v_data);

END;

/

Enjoy.

 If you are looking to use AI Transcription during your job interview to get help from LLM, then this video is going to help you immensely.

In the rapidly evolving landscape of recruitment, the integration of Artificial Intelligence (AI) tools has been a game-changer, especially AI transcription services. These services, which convert speech into a written or electronic text document, are now being increasingly utilized during candidate interviews. This innovative approach is not just about efficiency; it's about transforming the interview process into a more insightful, fair, and strategic component of talent acquisition. Let's delve into the merits of using AI transcription in candidate interviews and how it is reshaping the future of hiring.

Enhanced Accuracy and Objectivity

One of the paramount benefits of using AI transcription is the enhanced accuracy and objectivity it brings to the recruitment process. Traditional note-taking methods during interviews are inherently flawed, prone to human error, and can be biased. AI transcription, on the other hand, captures every word as it is spoken, ensuring that nothing is missed or misinterpreted. This verbatim record allows hiring managers to review the interview with a fine-tooth comb, ensuring that decisions are made based on what was actually said, rather than on recollection or interpretation.

Improved Candidate Experience

The candidate experience is crucial in attracting top talent. AI transcription contributes to a more engaging and less intimidating interview environment. Candidates can speak freely, knowing that their responses are being accurately recorded. This level of precision in capturing responses allows candidates to feel heard and valued, improving their overall experience and perception of the company. Furthermore, the efficiency of AI transcription speeds up the interview process, enabling quicker feedback and follow-ups, which candidates appreciate.

Accessibility and Inclusivity

AI transcription makes interviews more accessible and inclusive, breaking down barriers for candidates who may be deaf or hard of hearing. By providing real-time transcription, these candidates can participate fully in the interview process, ensuring equal opportunities for all. This commitment to inclusivity not only enhances the company's reputation but also widens the talent pool by making it more diverse.

Streamlined Analysis and Decision Making

The comprehensive data captured by AI transcription allows for a more streamlined analysis of candidates' responses. Hiring teams can utilize this data to identify patterns, evaluate key competencies, and make more informed decisions. AI tools can also highlight specific skills and qualifications mentioned during the interview, making it easier to assess candidates against job requirements. This data-driven approach reduces the likelihood of cognitive biases affecting hiring decisions and helps in selecting the best fit for the role.

Enhanced Collaboration Among Hiring Teams

AI transcription facilitates better collaboration among hiring team members, especially in scenarios where not all members can attend every interview. The transcribed interviews can be easily shared, allowing team members to review and provide input at their convenience. This collaborative approach ensures a more comprehensive evaluation of candidates, drawing on diverse perspectives within the team.

Legal Compliance and Record-Keeping

In today's litigious environment, maintaining accurate records of candidate interviews can be crucial for legal compliance. AI transcription provides an exact record of the conversation, which can be invaluable in defending against claims of discrimination or bias. This level of documentation ensures that the hiring process is transparent and defensible, safeguarding the company against potential legal challenges.

Conclusion

The adoption of AI transcription in candidate interviews is more than just a nod to technological advancement; it's a strategic move towards making the recruitment process more accurate, fair, and inclusive. By enhancing the accuracy of interviews, improving candidate experience, and streamlining decision-making, AI transcription is setting a new standard in talent acquisition. As companies continue to seek competitive advantages in attracting top talent, the use of AI in recruitment processes, especially transcription, is likely to become not just common but essential.

 Unlock the power of Retrieval-Augmented Generation (RAG) locally using the Gemma model with our detailed step-by-step tutorial. Learn how to enhance your projects by integrating RAG for insightful document processing and AI-driven content generation. Perfect for developers, data scientists, and AI enthusiasts eager to leverage advanced NLP techniques on their own documents. No prior RAG experience required!

Commands Used:

pip install -U "transformers==4.38.1" --upgrade

pip install -q pypdf

pip install -q python-dotenv

pip install  llama-index==0.10.12

pip install -q gradio

pip install einops

pip install accelerate

pip install llama-index-llms-huggingface

pip install llama-index-embeddings-fastembed

pip install fastembed

import logging

import sys

logging.basicConfig(stream=sys.stdout, level=logging.INFO)

logging.getLogger().addHandler(logging.StreamHandler(stream=sys.stdout))

from llama_index.core import VectorStoreIndex, SimpleDirectoryReader

from llama_index.llms.huggingface import HuggingFaceLLM

from llama_index.core import Settings

documents = SimpleDirectoryReader("/home/ubuntu/pdfs").load_data()

from llama_index.embeddings.fastembed import FastEmbedEmbedding

embed_model = FastEmbedEmbedding(model_name="BAAI/bge-small-en-v1.5")

Settings.embed_model = embed_model

Settings.chunk_size = 512

from llama_index.core import PromptTemplate

system_prompt = "You are a Q&A assistant. Your goal is to answer questions as accurately as possible based on the instructions and context provided."

query_wrapper_prompt = PromptTemplate("<|USER|>{query_str}<|ASSISTANT|>")

from huggingface_hub.hf_api import HfFolder 

HfFolder.save_token('<your huggingface token from huggingface.co>')

import torch

llm = HuggingFaceLLM(

    context_window=8192,

    max_new_tokens=256,

    generate_kwargs={"temperature": 0.7, "do_sample": False},

    system_prompt=system_prompt,

    query_wrapper_prompt=query_wrapper_prompt,

    tokenizer_name="google/gemma-7b-it",

    model_name="google/gemma-7b-it",

    device_map="auto",

    tokenizer_kwargs={"max_length": 4096},

    model_kwargs={"torch_dtype": torch.float16}

)

Settings.llm = llm

Settings.chunk_size = 512

index = VectorStoreIndex.from_documents(documents)

query_engine = index.as_query_engine()

def predict(input, history):

  response = query_engine.query(input)

  return str(response)

  

import gradio as gr

gr.ChatInterface(predict).launch(share=True)

  

 Oracle Database seamlessly integrates JSON (JavaScript Object Notation) with its robust RDBMS features, offering full support for transactions, indexing, declarative querying, and views among other functionalities.

JSON is a lightweight data interchange format widely employed for representing structured data due to its human-friendly readability and machine-friendly parsing capabilities. It utilizes key-value pairs enclosed in curly braces {}. Each key is a string followed by a colon :, and its value can be a string, number, boolean, array, object, or null. JSON arrays are ordered lists enclosed in square brackets [], while objects represent collections of key-value pairs. This versatile format finds extensive use in web development, APIs, and system-to-system data exchanges owing to its simplicity, adaptability, and clarity—a near subset of JavaScript's object literal notation.

When leveraging JSON within Oracle Database, the typical workflow involves:

Step 1: Creating a JSON-Enabled Table:

CREATE TABLE json_purchase_orders

  (id VARCHAR2(32) NOT NULL PRIMARY KEY,

   date_loaded TIMESTAMP WITH TIME ZONE,

   po_document JSON);

Step 2: Inserting JSON Data into the Table:

INSERT INTO json_purchase_orders

  VALUES (SYS_GUID(),

          TO_DATE('30-DEC-2014'),

          '{"PONumber": 1600,

            "Reference": "ABULL-20140421",

            "Requestor": "Alexis Bull",

            "User": "ABULL",

            "CostCenter": "A50",

            "ShippingInstructions": {...},

            "SpecialInstructions": null,

            "AllowPartialShipment": true,

            "LineItems": [...]}');

Step 3: Querying JSON Data from the Table:

SELECT po.po_document.PONumber

FROM json_purchase_orders po;

By leveraging Oracle Database's native JSON support, developers gain the flexibility to seamlessly integrate JSON data within the relational model, enabling efficient storage, retrieval, and manipulation of JSON documents alongside traditional SQL operations. This unified approach empowers enterprises to harness the best of both worlds—structured data management and JSON flexibility—within a single, powerful database environment.

 In this video, you will learn how to deploy a LLM based application intro production by using Amazon Bedrock, Amazon Transcribe to summarize audio files with ASR model, Titan. 

Code:

This video shows how to install and manage Gemma LLM with Keras. Gemma is a family of lightweight, state-of-the-art open models from Google, built from the same research and technology used to create the Gemini models.

Code:

 

Oracle GoldenGate is a comprehensive software product that enables real-time data integration, replication, and synchronization across heterogeneous systems. It supports continuous data streaming with minimal latency and offers high availability and data integrity. GoldenGate facilitates data replication between different database platforms, including Oracle Database, Microsoft SQL Server, MySQL, and others, allowing organizations to achieve real-time data integration for analytics, reporting, disaster recovery, and distributed data environments. Its capabilities include data transformation, filtering, conflict resolution, and support for complex replication topologies, making it a powerful solution for maintaining synchronized data across diverse IT landscapes.

This script checks if Oracle Golden Gate process is running or not and give the exact starting point to catch up:

#!/bin/bash

. /etc/profile

#. /home/oracle/.bash_profile 

YYMMDD=`date +%Y%m%d`

LOG_DIR=/home/oracle/GGscript/log

WRK_DIR=/home/oracle/GGscript/work

LOG_FL=${LOG_DIR}/${ORACLE_SID}_gg_chk.log

WRK_FL=${WRK_DIR}/gg_chk_${ORACLE_SID}_gg_chk.wrk

BAD_FL=${WRK_DIR}/gg_chk_${ORACLE_SID}_gg_chk.bad

export ORACLE_HOME=/u01/app/oracle/product/11.2.0.3/db_home1

export PATH=$ORACLE_HOME/bin:$PATH/bin

export LD_LIBRARY_PATH=$ORACLE_HOME/lib:/usr/local/lib

DBA_MAIL_ADDR='ADD_EMAIL@HERE.com'

HOSTNAME=`uname -n | tr '[a-z]' '[A-Z]'`

{

echo "###########################################"

echo "GG Check Up `basename $0` started t `date`"

#check if maint flag is on

if [ -f /home/oracle/work/${ORACLE_SID}_maintenance.on ] ; then

   echo "Maintenance flag is on .... stop "

   exit 0

fi

### Check GG processes

cd /ggs

echo "info all" | /ggs/ggsci | sed -e '1,/^Program/d' -e '/^GGSCI/,$d' -e '/^ *$/d' | tee $WRK_FL

echo " "

cat $WRK_FL | grep -v 'RUNNING'| grep -iv 'test' | tee $BAD_FL

non_run_cnt=0

if [ -f $BAD_FL ] ; then

   non_run_cnt=`wc -l $BAD_FL |awk '{print $1}'`

fi

echo "GG Process Not Running: $non_run_cnt "

if [[ $non_run_cnt -gt 0 ]] ; then

   echo "\nEmail support team ($DBA_MAIL_ADDR) about failed processes "

   cat $BAD_FL

   mailx -s "$HOSTNAME : ALERT: One or more Goldengate Process in ${ORACLE_SID} are down ($non_run_cnt)!!!" $DBA_MAIL_ADDR < $BAD_FL

fi

rm -f $WRK_FL $BAD_FL  $LAGWRK_FL $LAG_FL $LAG_LST

echo "GG Check Up `basename $0` finished at `date`"

echo "###########################################"

} 2>&1 | tee $LOG_FL