Express YourSelf !

• Background on NLP and tasks

NLP Tasks

1. Classification

- Sentimental analysis :  Amazon reviews, IMDB critiques, Twitter

- Intent detection

- Language detection

- Topic modeling

2. "Multi"-Classification

- Part of speech tagging

- Named entity recognition (NER): Dataset = annotated Reuters newspaper (CONLL-2003, CONLL+)

- Dependency parsing

- Constituency parsing

3. Generation

- Machine translation: Dataset = WMT'14

- Question answering

- Summarization

- Text generation

History of LLM

1980 RNN

1997 LSTM (Theoretical Foundation) 

2013 Word2Vec

2020s LLM

• Tokenization

1. Arbitrary (n/a)

2. Word (multiple tokens with similar meanings need same embedding, so Word variations not handled)  

3. sub-word : focus on common root. Increase sequence length. Tokenization more complex

4. character level: can correct mis-spelled word & CasINg. Sequence length is much longer. No OOV

• Embeddings

Word (Token) Representation by vector

OHE = One Hot Encoding

cosine similarity 

• Word2vec, RNN, LSTM

1. Word2Vec

It is ANN with proxy-task

1. CBOW: Continuous Bag of Words. You predict the target word 

2. Skip-gram : You take the target word and predict words around it

Word order does not matter

Embeddings is not context aware

Dimension size example 768

Special token to indicate "end of sequence" 

2. RNN Recurrent Neural Network

Connection forms a temporal sequence

H = Hidden state = A = Activation Vector = Context Vector. 

RNN is used for all 3 NLP tasks

1. Classification

2. "Multi"-Classification

3. Generation

RNN is keep forgetting the past. This phenomena is called "vanishing gradient"

Word order matters in RNN

3. LSTM = Long short-term memory

1. hidden state

2. cell state

• Attention mechanism

Attention tries to have a direct link between next word that we are predicting and something from the past. 

"self-attention" is main principle of "Attention is all you need" 2017 paper

"self-attention" = Instead of sequential, let direct connection with all part of text at once. 

Concept of Query, Key and Value

We compare Q to K. How they are similar and then take corresponding value

Softmax converts unnormalized network output into probability for different class such that value is [0,1] and sum is 1. 

Formula – Given a query Q, we want to know which key K the query should pay "attention" to with respect to the associated value V. 

attention = softmax ( Q * K ^ T / Sqrt (dimension of K) ) * V

There are three attention layers

1. Attention layer at encoder to compute embeddings from input

2. Decoder-decoder attention OR self-attention layer in decoder, It is is masked, because it only look at those token that are translated. It determines: what other token of output sentence is useful to predict next token. 

3. cross-attention layer : expressed as function of what is seen in input. Last part of encoder. it is fetch to decoder. 

We have direct link to all token. So order words does not matter. (unlike RNN).  So we have Position Encoding: to inform position of word in sequence. 

BOS Token: Beginning of Sequence. 

EOS Token: End of Sequence

• Transformer architecture

Self-attention is achieved by transformer = encoder and decoder

1. Encoder computes meaningful embedding from input text. We have N such encoders. Input layer generates position aware embedding matrix with size d = model size and length = length of input sequence = n

Encoder projects input sequence on 3 spaces Wk, Wq and Wv. so model learns. 

attention = softmax ( Q * K ^ T / Sqrt (dimension of K) ) * V

Projecting on Wq gives a matrix where each row represents a given query Q. So we get matrix Wo that is project back to original dimension of embedding. 

K^T is each column represents key of each token. 

When we multiple K^T and Wq, Each row represents projection of query over each key and then get probability distribution. 

Now multiple with matrix V 

This is self-attention mechanism. means compute representation of each token as function of other tokens. it is done by attention layer. 

Multi-Head Attention (MHA) means this computation is done in different way. So model can learn 

- different representation

- different projections

so all token of input text attend each other. 

It is masked self-attention layer. 

A Multi-Head Attention (MHA) layer performs attention computations across multiple heads, then projects the result in the output space.

2. FFNN (Feed Forward Neural Network) : so model learn another kind of projection

so we get rich representation of input token

In LLM, hidden layer has higher dimension. So model has enough degree of freedom to learn useful representation. 

3. output is for decoder

It takes Q from output. 

K, V from encoder. 

we have N decoders. 

New Terms

• Perplexity is an evaluation matrix for machine translation. It quantifies how 'surprised' the model is to see some words together. Lower is better. 
• OOV = out of vocabulary
• RNN is keep forgetting the past. This phenomena is called "vanishing gradient"

• Label Smoothing Purpose

  - prevent overfitting

  - introduce noise

  - let model be little unsure about prediction. 

  It improves accuracy and BLEU score of translation.

References

https://cme295.stanford.edu/

Syllabus : https://cme295.stanford.edu/syllabus/

CheatSheet 

https://cme295.stanford.edu/cheatsheet/ 

https://github.com/afshinea/stanford-cme-295-transformers-large-language-models/tree/main/en 

https://www.youtube.com/watch?v=Ub3GoFaUcds

Text Book Super Study Guides

------------------------------------------------------

Some more relevant stuff: 

Each layer has 

1. Attention and 

2. Fast Forward

Between two layers we have high dimension 'hidden state vector' in activation space. 

LLM encodes concepts as distributed patterns accross layers = Superposition. 

Antropic has series of papers on superposition and monosemanticity

https://www.youtube.com/watch?v=F2jd5WuT-zg

https://www.neuronpedia.org

https://huggingface.co/collections/dlouapre/sparse-auto-encoders-saes-for-mechanistic-interpretability

https://huggingface.co/spaces/dlouapre/eiffel-tower-llama

------------------------------------------------------------

On January 18, 2026 BAPS Banglore temple hosted IT convention event from morning to evening. More than 375 participants. 

1.  "Changing Trends of AI in technologies" by Prof. Rahul De

Prof. Rahul De' is founder and CEO of https://www.memoricai.in/ He provided nice academic insight, history of AI, present state and future. AI is about inference and inference is predications, classification and generative output

Human brain has 80 to 86 billion neurons (cells). 

Evolution of AI

1. ANI : Artificial Narrow Intelligence 

2. GNI : General Narrow Intelligence 

3. ANI : Super Narrow Intelligence 

In 1966 a professor Joseph Weizenbaum at MIT developed first chatbot by name ELIZA. It acts like a Rogerian psychotherapist. People like it so much. Later on, we had to convince, that it is not a real person. It is just a computer program that simulates human conversation, through pattern matching and keyword substitution.

Late in 1980 John Searle did "Chinese Room Experiment". Here, a non-Chinese speaker in a room, just manipulate Chinese symbols manually and produce fluent responses without understanding the language.  It proves that just through syntax (rule-following) alone, computer cannot achieve semantics (genuine understanding or consciousness). 

Probably that is why today, GenAI has caveats like hallucinations, jail breaking, bias, privacy violations and unfair responses. In the context of bias, he mentioned about recent movie "Human in the loop"  Available on Netflix. "An indigenous woman works as an AI data-labeler after returning to her village with her children, but soon questions the human bias in machine learning."

He shared some statistics

• - 2.5 billion prompts are handled by ChatGPT alone in a year. 
• - 2.4 million models are present at hugging face
• - 50 billion USD are spent for AI in year 2025. 
• - 95% firms fail in GenAI adoption. 
• - We achieved 15% improvement by GenAI

The above numbers raised serious questions that does spending behind GenAI is worth? 

He mentioned few books and categorised all AI adopters in four groups. Boomers (books by Ray Kurzweil), Doomers , Skeptics and Critics "Empire of AI: Dreams and Nightmares in Sam Altman's OpenAI"

2. panel discussion on "Ways to overcome Challenges in IT"

One of the discussion point was about reducing 95% failure rate in 2026

We need: 

• - Automated task workflow
• - Cross functional aggregation across various departments
• - Collaboration between AI and human
• - Orchestration for AI in day-today work

How can you fail? Your effort (to adapt AI, to retain job etc) can fail. In fact the definition for failure came after industry revolution. During the last century, "Productivity" was more in focus due to industry revolution. 

Other points: 

Former OpenAI co-founder Ilya Sutskever has indicated that simply scaling up to 1 trillion parameters model will not improve AI capability further. 

May be, personalised AI will be next thing

We are humans so 

• - we are always optimistic, we have hope. 
• - We have ability to adapt.
• - We are creator of AI so we are smarter than AI 

We shall remove the fear that we need to learn everything. Yes, we shall learn something new everyday and take its now on NotebookLM. Internet is flooded with many buzzword about AI. We need to separate signal from noise. 

Now learning is not same as degree earning. 

Now, we need to be aware about all domain. That knowledge shall not be gain by asking ChatGPT. 

Few points were discussed about parenting: We shall have 30 min of productive arguments with kids. We will learn AI from the end-user. We shall enable parental control for Internet, OTT. While using AI, be skeptic. You are interacting with product. So product has market, company / organisation behind it, that want to earn profit. AI product is not your friend. More we use LLM, that much brain is unused and lost.  

• - Humans are not means
• - Respect life in people

About firing due to AI. 

• * If software engineers consider themselves as coder then AI will replace them. They shall consider themselves as problem solver. 
• * On lighter notes: Pujya Aksharatit Swami mentioned that we SADHUs are easier to get replace. Chat with AI is available round the clock. 
• * On lighter notes: Pujya Aksharatit Swami mentioned 

येषाम् न विद्या न तपः न दानम् न ज्ञानम् न शीलम् न गुणः न धर्मः ।।

ते मृत्यु-लोके भुवि भार-भूता मनुष्यरूपेण मृगाः चरन्ति ।

It means: Those who possess no knowledge (Vidya), no penance (Tapa), no charity (Dana), no wisdom (Jnana), no good character (Sheela), no virtues (Guna), and no righteousness (Dharma), are a burden to the earth. Although they look like humans, they roam the earth like animals in human form. 

Now this sloka is applicable for knowledge of AI also :-) 

During the panel discussion, the floor is open for everyone to ask question via WhatsApp group, that was flooded with many questions. 

3. Networking

The audience was divided into many groups. The participants had a round of introductions within group. There was engaging quiz, where all groups participated and the group leader responded to questions on behalf of group. 

We had delicious vegetarian, SAATTVIK food. Now some key take away points from post lunch sessions.

4. Work-Life Integration & Wellbeing

“In God we trust. All others must bring data.” - W. Edwards Deming

Here are some shocking data points

• 83% of Indian IT professional are burnout
• 73% of European IT professional are burnout
• 72% are working beyond limit

The data source is ISACA (Information Systems Audit and Control Association)

5 pillars of work life integration

1. Know your why?

What is purpose? It brings impact, mastery and autonomy. 

Our health and family should tune to work. 

2. Design your system

Define boundaries so you can protect capacity

Bring rhythms by frequent breaks. 

3. Recovery and build resilience practices

3.1 Mindful ness and reflection. Moment to moment non-reactive, non-judgmental awareness is mindful ness. 

3.2 physical movement

3.3 social connections

4. Align your environment

5. Seek help early

Something about sleep

- Sleep is non-negotiable.

- Sleep hygiene and nutrition  

- Stop all screens 60 to 90 min before going to bed. It is digital sunset. 

- use eye mask while sleeping. 

Next topic was NSANE

Nutrition

Screen Time

Automatic health. He mentioned about book : "Atomic Habit" 

Notice Signal

Engage with real people

Remember 3 truths

1. Mind = body. Means if body is unhealthy, means mind is unhealthy. Body can be healthy by making healthy mind

2. Work family balance is important

3. It is still not too late. 

Ask your wife, what she expect about husband? 

- Rich husband?

- Healthy husband? 

- Rich but unhealthy husband? 

- burn out person as husband? 

3. Personal/Spiritual Fire Chat with Pujya Santo

Along with other relevant questions and guidance by Pujya SAINTs, again IT layoff was discussed. The apparent reason is AI for layoff. The real reason can be poor performance of employee. Extra hiring happens during COVID phase, so now layoff is inevitable. जातस्य हि ध्रुवो मृत्युः Same way, if you have job, you may get fire. If not today then in future, at age of 62 years. Even after 10 years, the present software application has no values. This is also as per SANKHYA philosophy. It inspires us to make better documentation of product. 

Summary

1. Be happy

2. Worship God. 

બી.એ.પી.એસ. પ્રકાશ એપ

પ્રકાશ વિશેષ :

બેંગલુરુમાં આઇ.ટી. પ્રોફેશનલ્સ માટે યોજાયો વિશેષ સમારોહ

વધુ માહિતી માટે નીચેની લિંક પર ક્લિક કરો.

https://bapsprakash.in/deeplink?data=QsfHnR6PaUp%2FlihVajHZKg%3D%3D%3AWm7u3d6zy%2FylpRGzMz6xFENQBWhBhVioEwCNJCh3MI8%3D

Disclaimer: The author had put best effort to capture all the points, as per his understanding. It may or may not reflect exact intention of the speaker. So any corrections are welcome. This article is not verbatim 

This video is about MCP, ACP and A2A but later 11+ more protocols were found

MCP : Design goal is not Agent to Agent communication. But now it is happening. 

- MCP is also wroking on Agent registry. 

- It is different from A2A (1) No async message. (2) No renegotation. 

- It is most adopted protocol. 

- MCP has conquer world like React.js

These protocols are about : Discover, communicate, authenticate. over Internet

1. Inter-Agent protocol

1.1. Robot-Agent: CrowdES, SPPs (Saptial Population Protocol) 

1.2. Human Computer: LOKA, PXP

1.3. System-Agent: Agent Protocol, LMOS (Large Model Operating System)

2. Context Oriented

2.1 MCP

Few Protocols

• Agents.JSON: File has API documentation for Agent. It is in OpenAPI format
• ANP: Agent Network Protocl. DID Distributed Identity (Blockchain-based)
• AITP: Blockchain. interaction cost. 
• ACP (Agent Connect Protocol) by Cisco. Like A2A except it describe how to host and launch agent
• ACP (Agent Communication Protocol) by IBM. like MCP. It has registry = distributed Database of all agents. It is fork of MCP
• Agora: Natural language protocol. protocol upgrade itself. 

All protocols lack

1. Regsitry

2. Authorizatoin

3. Reputation

https://www.youtube.com/watch?v=kqB_xML1SfA

=============================

OAuth is bunch of specs

OAuth2.0 = RFC6749 (2012) + RFC6750

OAuth2.1 = Authorization code + PKCE, Client credentials (may not useful), token in HTTP header, token in POST form body. Refer MCP Github issue 830

OpenID Connect is to identify user. It provides ID token

RFC 9728 : single URL for MCP server. It points to file that define meta data of server. it has auth server meta data as per RFC 8414. 

RFC 7591 : Dynamic client registration to get client ID and Client secret. 

aaronpk.com/mcp

oauth.wtf

oauth.net

https://www.youtube.com/watch?v=mYKMwZcGynw

==========================================

Keycloak and SPIRE

CNCF IAM Whitepaper. Yet to be published

5 principals

1. mTLS

2. OAuth2 token exchange RFC 8693

3. OIDC client authentication with SPIFEE SVID

4. OAuth2 token validation

5. PDP based Authorization decision. 

PDP, PAP, PIP, PEP

• PAP (Policy Administration Point): The user interface/management layer where policies (rules) are created, stored, and managed centrally.
• PDP (Policy Decision Point): The "brain" that evaluates a user's request against policies from the PAP, using attributes from PIP, to return an "Allow" or "Deny" decision.
• PIP (Policy Information Point): Fetches necessary attributes (like user roles, data from databases/LDAP) needed by the PDP to make a decision.
• PEP (Policy Enforcement Point): Sits in front of the protected resource, intercepts requests, sends them to the PDP for a decision, and enforces that decision. 

1. A user requests a resource (e.g., "view my profile").

2. The PEP intercepts the request and the OAuth access token.

3. The PEP sends the request details (user context from token, resource, action) to the PDP.

4. The PDP queries the PIP (e.g., a database) for attributes like user's department.

5. The PDP evaluates policies (managed by PAP) and returns "Permit/Deny" to the PEP.

6. The PEP allows or blocks the user's request based on the decision. 

https://www.youtube.com/watch?v=S6qF0N5D1tM

10 days Spoken Sanskrit workshops (Sanskrit Sambhaasan Shibir) held from July14 2025 to July 23 2025 at BEML Balaji Temple and at Skylark Arcadia Bangalore, by Samskrita Bharati Marathahalli bhaga, Bangalore, India. 

Spoken Sanskrit workshops - closing ceremony on 26th July 2025.

Sanskrit speech by chief guest Shri Manish Panchmatia

Greetings!

A warm welcome to everyone present—teachers, volunteers from Samskrita Bharati, students, and all those passionate about Sanskrit. After bowing down to each of you, I am honored to begin my speech.

First, let me ask: “How was the Spoken Sanskrit Workshop?” Was it “Good”? “Very good”? I’m certain it was a positive experience. Whenever Samskrita Bharati organizes a workshop, the teachers impart their knowledge with dedication, don’t they? Over the past ten days, each one of you demonstrated incredible determination to learn Sanskrit. No matter what, you attended all the classes and gave your best effort. Kudos to you all for your commitment and enthusiasm. You sang songs in Sanskrit, offered prayers, performed dramas, and narrated stories—all in Sanskrit. I am truly impressed by your efforts and progress.

As you may know, NASA called Sanskrit the most suitable language for computers. I won’t repeat those facts, but I do want to share a story about Sanskrit’s vast vocabulary. Just now, we sang about great poets—Vyasa, Bhasa, Kalidasa, Banabhatta—yet there is another renowned poet: Dandi, who composed “Dasakumaracharitam,” the story of ten princes. One prince, wounded in the lower lip, could not pronounce certain sounds: "pa," "fa," "ba," "bha," and "ma." Dandi skillfully chose synonymous words without those sounds for all of that prince’s dialogue. This was possible only because Sanskrit’s vocabulary is so rich.

Sanskrit grammar is exceptionally precise; in fact, the very meaning of the word “Sanskrit” is “well-formed” or “perfect.” Sanskrit has maintained its grammar rules since Vedic times—pronunciation, grammar, language rules—all unchanged. Even though new inventions like mobile phones have appeared, in Sanskrit, we create new words effortlessly. “Mobile phones” in Sanskrit is “Jangam Door-Vani”—a moving telephone. Artificial intelligence becomes “Krutrim Buddhi.” Thanks to Sanskrit’s structure, new words can always be formed. This habit of following rules brings discipline to our lives. 

We all are Indians. Sanskrit is deeply imprinted in our consciousness. Just as our childhood photos make us happy, speaking Sanskrit awakens ancient impressions within us—it fills the soul with joy. Throughout these ten days, you learned Sanskrit joyfully, supported by equally joyful teachers.

How do we gain these benefits? By giving your 100% effort to learning Sanskrit. What does 100% effort mean? Here is another story from the Mahabharata. Karna is renowned as the greatest giver. Once, Arjuna asked Krishna, "Why is Karna more famous for charity than my brother Yudhishthira?" Krishna replied, “Let’s see for ourselves.” Disguised as BRAHMINs, Arjuna and Krishna visited Yudhishthira and requested sandalwood for a yagna. Yudhishthir responded, "OK". Then, what did Yudhishthir do? He searched. Remember the story of the crow. The crow was searching for water. Here Yudhishthir is searching for sandalwood. He looked here . He looked there. Right side. Left side. Front side. Back side. He looked for Sandalwood everywhere. The sandalwood is nowhere. Then he went outside. Looked at the garden. He searched everywhere. The sandalwood is nowhere. He returned back and told BRAHMIN: "Sorry sir. The sandalwood is nowhere. I will surely make a little more effort to arrange for sandalwood. Please come tomorrow. I will give." BHRAMIN said, "OK".  Then, they visited Karna with the same request. Karna eagerly searched, and when he could not find sandalwood, he noticed his door was made of sandalwood. Without hesitation, he dismantled it and gave it to them. True charity is giving with 100% effort. It is the same with learning Sanskrit—when you give your wholehearted effort, you will gain both knowledge and discipline.

Further, speaking Sanskrit even helps with breathing exercises—“ma” and “ha” are frequently used, which naturally leads to “pranayama.”

This spoken Sanskrit workshop is just the beginning. Explore further—enroll in correspondence Sanskrit learning courses, Gita Sopanam etc. There are many Sanskrit books here, for sale. Buy them, read them and then, become Sanskrit teachers yourselves! Spread and promote Sanskrit to others just as your teachers did for you. You all know Swami Vivekananda. Right? "Yes". On 4th July, it was his death anniversary. On the last day of life, he taught Sanskrit to the students. You all know that? It is in his biography. Let us honor this legacy by teaching and learning Sanskrit.

Having spoken much about Sanskrit, let me turn briefly to our IT professionals. When we hear the word “language,” we often think of C, C++, Java, Python, NodeJS, ReactJS, and so on. Recently, I attended a "PyKrit" (Python + Sanskrit) workshop at Aksharam, Samskrita Bharati’s center.  This workshop was not for IT people. It was for Sanskrit scholars. I saw Sanskrit scholars name Python functions in Sanskrit, such as “YANA SANDHI” (a grammar topic). We truly can create software for Sanskrit grammar, using modern programming languages like Python, and even coding in Sanskrit.

Lastly, my wish: While we use AI/GenAI tools like ChatGPT, we mostly interact in English. Wouldn’t it be wonderful to have a large language model (LLM) in Sanskrit? Imagine asking questions in Sanskrit and receiving answers, back in Sanskrit, from AI tools, such as ChatGPT. That is my hope for the future.

Best wishes to everyone, and thank you all.

For AI application, we need automation of 

1. Data preparation
2. model tuning
3. Deployment
4. Maintenance and 
5. Monitoring

• Managing Dependency adds complexity. 

E2E workflow for LLM based application. 

MLOps framework

1. data ingestion

2. data validation

3. data transformation

4. model

5. model analysis

6. serving model

7. logging. 

LLM System Design

boarder design of E2E app including front end, back end, data engineering etc. 

Chain multiple LLMs together

* Grounding : provides additional information/fact with prompt to LLM. 

* Track History. how it works past. 

LLM App

User input->Preprocessing->grounding->prompt goes to LLM model->LLM Response->Grounding->Post processing + Responsible AI->Final output to user.

Model Customization

1. Data Prep

2. Model Tuning

3. Evaluate

It is iterative process

LLMOps Pipeline (Simplified)

1. Data Preparation and versioning (for training data)

2. Supervised tuning (pipeline) 

3. Artifact = config and workflow : are generated. 

- config = config for workflow

E.g. 

Which data set to use

- Workflow = steps 

4. Pipeline execution

5. deploy LLM 

6. Prompting and predictions

7. Responsible AI

Orchestration = 1 + 2 . Orchestration : What is first, then next step and further next step. sequence of step assurance. 

Automation = 4 + 5

Fine Tuning Data Model using Instructions (Hint)

1. rules

2. step by step

3. procedure

4. example

File formats

1. JSONL: JSON Line. Human readable. For small and medium size dataset. 

2. TFRecord 

3. Parquet for large and complex dataset. 

MLOps Workflow for LLM

1. Apache Airflow

2. KubeFlow

DSL = Domain Specific Language

Decorator 

@dls.component

@dls.pipeline

Next compiler will generate YAML file for pipeline

YAML file has

- components

- deploymentSpec

Pipeline can be run on

- K8s

- Vertex AI pipeline execute pipeline in serverless enviornment

PipelineJob takes inputs

1. Template Path: pipline.yaml

2. Display name

3. Parameters

4. Location: Data center

5. pipeline root: temp file location

Open Source Pipeline

https://us-kfp.pkg.dev/ml-pipeline/large-language-model-pipelines/tune-large-model/v2.0.0

Deployment

Batch and REST

1. Batch. E.g. customer review. Not real time. 

2. REST API e.g. chat. More like teal time library. 

* pprint is library to format 

LLM provides output and 'safetyAttributes'

- blocked

* We can find citation also from output of LLM

===========

vertexAI SDK

https://cloud.google.com/vertex-ai

BigQuery 

https://cloud.google.com/bigquery

sklearn

To decide data 80-20% for training and evaluation. 

Building AI/ML apps in Python with BigQuery DataFrames | Google Cloud Blog

===========