Natural Language Processing , LLM with Python

In this video I will be explain about Stemming and Lemmatization in the context of Natural Language Processing.

I will also explain how you can do stemming and lemmatization using NLTK and spaCy packages in Python.

The code is present here: https://github.com/rsreetech/StemmingAndLemmatization

Relevant Links:

https://nlp.stanford.edu/IR-book/html/htmledition/stemming-and-lemmatization-1.html

https://en.wikipedia.org/wiki/Stemming

https://nlp.stanford.edu/IR-book/html/htmledition/stemming-and-lemmatization-1.html

https://en.wikipedia.org/wiki/Lemmatisation

In this video i will be explaining concepts of Bag of words, Term frequency- Inverse Document Frequency, Cosine similarity in the context of Natural Language Processing. I will also explain how to implement these concepts in python.I have also shared the code on Github

Github

https://github.com/rsreetech/Tf-idf

Links: TfidfVectorizer : https://scikit-learn.org/stable/modules/generated/sklearn.feature_extraction.text.TfidfVectorizer.html

NLTK: https://github.com/nltk/nltk

Cosine Similarity : https://scikit-learn.org/stable/modules/generated/sklearn.metrics.pairwise.cosine_similarity.html

In this video i will be explaining about exploratory data analysis and visualization on text data.I will be using the Amazon fine food reviews dataset from Kaggle(https://www.kaggle.com/snap/amazon-fine-food-reviews). I will be using libraries like seaborn, wordcloud to explore and visualize the text dataset

The code is present here: https://github.com/rsreetech/TextDatasetEDAVisualization

In this video I will be explaining what is Named Entity Recognition(NER) in the context of Natural Language Processing. We also look into how we can use spaCy library in python to perform named entity recognition.

spaCy https://spacy.io/

In this video I will explain a method for clustering the CORD-19 dataset. CORD-19 is a free resource of tens of thousands of scholarly articles about COVID-19, SARS-CoV-2, and related coronaviruses for use by the global research community. I have used K-means clustering on the abstracts of articles in the CORD-19 dataset.I have then generated wordcloud on the titles of articles in each cluster to visualize the topics present in the cluster. The entire python code along with the clustered data is shared on github

In this video I will be explaining about Word2Vec and how to train your custom Word2Vec Model using genism in Python.

The code is present here: https://github.com/rsreetech/Word2VecCustomModel

Relevant Links:

Word2Vec papers :

https://arxiv.org/pdf/1301.3781.pdf

https://papers.nips.cc/paper/5021-distributed-representations-of-words-and-phrases-and-their-compositionality.pdf

Word2vec tutorial:

http://mccormickml.com/2016/04/19/word2vec-tutorial-the-skip-gram-model/

http://mccormickml.com/2017/01/11/word2vec-tutorial-part-2-negative-sampling/

gensim:  https://radimrehurek.com/gensim/models/word2vec.html

In this video I will demonstrate how we can implement text classification using Neural Network with TensorFlow 2.1 in python. I am specifically training a simple neural network based text classification model to classify sentiment of tweets . I will be using the tensorflow gpu version. The dataset is from the Tweet Sentiment Extraction challenge from Kaggle(https://www.kaggle.com/c/tweet-sentiment-extraction/overview) We would implement text classification using a simple neural network developed using Tensorflow on tweet data to classify tweets as "positive", "negative" or "neutral"

The code is present here: https://github.com/rsreetech/TextClassificationNNTensorflow

In this video I will demonstrate how we can implement text classification using Convolutional Neural Network with TensorFlow 2.1 in python. I am specifically training a simple convolutional neural network based text classification model to classify sentiment of tweets . I will be using the tensorflow gpu version The dataset is from the Tweet Sentiment Extraction challenge from Kaggle(https://www.kaggle.com/c/tweet-sentiment-extraction/overview) We would implement text classification using a simple neural network developed using Tensorflow on tweet data to classify tweets as "positive", "negative" or "neutral"

The code is present here: https://github.com/rsreetech/TextClassificationTensorFlowCNN

In this video I will demonstrate how we can implement text classification using LSTM using TensorFlow 2.1. I am specifically training a simple LSTM based text classification model to classify sentiment of Amazon Reviews . The dataset is from Amazon Review Data (2018) https://nijianmo.github.io/amazon/index.html. I here look at Cell Phones and Accessories review dataset from "Small" subsets for experimentation. I have pre-processed this dataset.

The script for pre-processing is present here: https://github.com/rsreetech/TextClassificationLSTMTensorFlow/blob/master/AmazonReviewsPreprocessing.ipynb

Let us perform sentiment classification by considering reviews which have rating 3(3,4,5) and above as having positive sentiment and reviews having ratings below 3(1,2) as negative sentiment. Let us develop a LSTM based classification model for sentiment classification on review data.

The code is present here: https://github.com/rsreetech/TextClassificationLSTMTensorFlow/

In this video I will be explaining my solution to Kaggle Natural Language Processing with Disaster Tweets (Getting Started Competition). I will be using BERT(Tensorflow) in the solution

The main purpose of the competition is Twitter has become an important communication channel in times of emergency. The ubiquitousness of smartphones enables people to announce an emergency they’re observing in real-time. Because of this, more agencies are interested in programmatically monitoring Twitter (i.e. disaster relief organizations and news agencies).

In this video I will be explaining how we can perfom text classification with spaCy v3.0 transformers library in Python.

spaCy v3.0 features all new transformer-based pipelines that bring spaCy’s accuracy right up to the current state-of-the-art.

The dataset is from the Tweet Sentiment Extraction challenge from Kaggle(https://www.kaggle.com/c/tweet-sentiment-extraction/overview)

We would perform text classification using spaCy on tweet data to classify tweets as "positive", "negative" or "neutral"

In this video I will explain how you can implement Text Classification on Custom Dataset using PyTorch and Torchtext. The dataset is from the Tweet Sentiment Extraction challenge from Kaggle(https://www.kaggle.com/c/tweet-sentiment-extraction/overview)

We would implement text classification using a simple embedding bag of words using PyTorch on tweet data to classify tweets as "positive","negative" or "neutral"

In this video I explain about Zero Shot Text Classification and Textual Entailment. Traditionally, zero-shot learning (ZSL) most often referred to a fairly specific type of task: learn a classifier on one set of labels and then evaluate on a different set of labels that the classifier has never seen before. Recently, especially in NLP, it's been used much more broadly to mean get a model to do something that it wasn't explicitly trained to do. A well-known example of this is in the GPT-2 paper where the authors evaluate a language model on downstream tasks like machine translation without fine-tuning on these tasks directly. I explain about how textual entailment is used in the context of zero shot learning. I also explain about Hugging face zero shot text classification pipeline with a demo

In this video I will be explaining about LDA Topic Modelling Explained and how to train build LDA topic model using genism in Python.

The code is present here:

https://github.com/rsreetech/LDATopicModelling Relevant Links:

LDA paper : https://ai.stanford.edu/~ang/papers/jair03-lda.pdf

LDA tutorial:

https://en.wikipedia.org/wiki/Latent_Dirichlet_allocation

https://www.machinelearningplus.com/nlp/topic-modeling-gensim-python/

In this video I demonstrate using a Kaggle notebook how topic modelling and semantic search can be done using Top2Vec python library. Top2Vec is an algorithm for topic modeling and semantic search. It automatically detects topics present in text and generates jointly embedded topic, document and word vectors. Once you train the Top2Vec model you can:

• Get number of detected topics.

• Get topics.

• Get topic sizes.

• Get hierarchichal topics.

• Search topics by keywords.

• Search documents by topic.

• Search documents by keywords.

• Find similar words.

• Find similar documents.

• Expose model with RESTful-Top2Vec

In this video I discuss about BERTopic. BERTopic is a topic modelling technique that leverages huggingface transformers and c-TF-IDF to create dense clusters allowing for easily interpretable topics whilst keeping important words in the topic descriptions.

I explain the functionality of BERTopic with a python demo on arxiv dataset

In this video I will explain about text summarization. I will be explaining about Text Rank Extractive Summarization Algorithm and I will also demonstrate how you can perform text summarization using spaCy (pytextrank) and genism in Python

The code is present here: https://github.com/rsreetech/TextSummarization/

Relevant Links:

TextRank paper : https://arxiv.org/abs/1602.03606

PageRank: https://en.wikipedia.org/wiki/PageRank

In this video I will explain about abstractive text summarization. I will be explaining about Google PEGASUS deep learning model and also demonstrate how you can perform abstractive text summarization using Hugging Face Transformers library in Python. This video will have 3 sections . The first section deals with abstractive text summarization concept, few pre-requisites like transformers, masked language model and rougeF1 metrics. The second section explains the PEGASUS sequence to sequence model for abstractive text summarization and the third section explains how you can use Pegasus on your text for abstractive text summarization using Hugging Face Transformers library in python

The code is present here: https://github.com/rsreetech/PegasusDemo/

Relevant Links:

Pegasus paper : https://arxiv.org/abs/1912.08777

Illustrated Transformer : http://jalammar.github.io/illustrated-transformer/

Illustrated BERT : http://jalammar.github.io/illustrated-bert/

ROGUE1-F1 : https://www.ccs.neu.edu/home/vip/teach/DMcourse/5_topicmodel_summ/notes_slides/What-is-ROUGE.pdf

In this video I demonstrate how Haystack can be used to build search systems. I use Haystack on MIMIC clinical data to build a simple search and question answering engine. You should have  access to MIMIC data if you want to try the demo

In this short video I discus about Question Answering with Hugging Face Transformers Pipeline in Python. Extractive Question Answering is the task of extracting an answer from a text given a question. I also explain how BERT transformer is used for the extractive question answering tasks

Relevant Links:

https://huggingface.co/docs/transformers/task_summary#question-answering

https://huggingface.co/course/chapter7/7?fw=tf

https://mccormickml.com/2020/03/10/question-answering-with-a-fine-tuned-BERT/

https://www.youtube.com/watch?v=l8ZYCvgGu0o

In this video I explain about Aspect Based Sentiment Analysis. It is the task is to classify the sentiment of potentially long texts for several aspects. I explain and give a demo of the python Aspect-Based-Sentiment-Analysis package

Relevant Links:

https://github.com/ScalaConsultants/Aspect-Based-Sentiment-Analysis

https://paperswithcode.com/task/aspect-based-sentiment-analysis

https://rafalrolczynski.com/2021/03/07/aspect-based-sentiment-analysis/

In this video I explain Prompt-based learning in natural language processing. In Prompt-based learning, instead of adapting pre-trained LMs to downstream tasks via objective engineering, downstream tasks are reformulated to look more like those solved during the original LM training with the help of a textual prompt. I explain what is prompt-based learning and a formal description of prompting. I refer to the excellent summary paper by Pengfei Liu et.all from CMU titled “Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processing”. (https://arxiv.org/abs/2107.13586 )

In this video I explain about Pre-trained Language Models for Prompt-based learning in natural language processing. In Prompt-based learning, instead of adapting pre-trained LMs to downstream tasks via objective engineering, downstream tasks are reformulated to look more like those solved during the original LM training with the help of a textual prompt. I explain what is prompt-based learning and a formal description of prompting. I refer to the excellent summary paper by Pengfei Liu et.all from CMU titled “Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processing”. (https://arxiv.org/abs/2107.13586 )

Pre-trained Language Models : In this paradigm, a model with a fixed architecture is pre-trained as a language model (LM),predicting the probability of observed textual data. Because the raw textual data necessary to train LMs is available in abundance, these LMs can be trained on large datasets, in the process learning robust general-purpose features of the language it is modeling. The above pre-trained LM will be then adapted to different downstream tasks by introducing additional parameters and fine-tuning them using task-specific objective functions.(from https://arxiv.org/abs/2107.13586 ))

In this video I explain Prompt-based learning in natural language processing. In Prompt-based learning, instead of adapting pre-trained LMs to downstream tasks via objective engineering, downstream tasks are reformulated to look more like those solved during the original LM training with the help of a textual prompt. I explain what is prompt engineering. I refer to the excellent summary paper by Pengfei Liu et.all from CMU titled “Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processing”. (https://arxiv.org/abs/2107.13586 )

In this video I explain some Interesting use cases of ChatGPT

I explain about GPT-4

The Assistants API allows you to build AI assistants within your own applications. An Assistant has instructions and can leverage models, tools, and knowledge to respond to user queries. The Assistants API currently supports three types of tools: Code Interpreter, Retrieval, and Function calling.

Using Assistants API I create a Data Scientist AI Assistant which can help you perform data scientist tasks on your dataset

https://colab.research.google.com/drive/15Ln2qkyxzQNd0HDWJfBcQsAj73bUr9k1?usp=sharing

https://www.kaggle.com/datasets/arnabchaki/data-science-salaries-2023

https://platform.openai.com/docs/assistants/overview

Text-to-SQL is a task in natural language processing (NLP) where the goal is to automatically generate SQL queries from natural language text. A more academic definition is to convert natural language problems in the database field into structured query languages ​​that can be executed in relational databases. Text-to-SQL aims to narrow the gap between database systems and non-technical users by automating the conversion of natural language queries into SQL commands. By utilizing the capabilities of large language models (LLMs), which enhances the Text-to-SQL process, thereby enhancing data processing efficiency and enabling a range of applications including automated data analysis, intelligent database services, and streamlined query responses.

Relevant Links:

https://www.kaggle.com/datasets/dillonmyrick/bike-store-sample-database

https://paperswithcode.com/task/text-to-sql

https://github.com/eosphoros-ai/Awesome-Text2SQL

Code: https://colab.research.google.com/drive/1MMX2S_jTxhvfDTu0uD7tHxmNY11FdYFZ?usp=sharing

In this video I explain about what is Retrieval Augmented Generation. Retrieval Augmented Generation (RAG) has emerged as one of the most popular techniques in AI over the past year, widely adopted by many AI teams. RAG refers to the process of supplementing a large language model (LLM) with additional information retrieved from elsewhere to improve the model’s responses.

https://www.anyscale.com/blog/a-comprehensive-guide-for-building-rag-based-llm-applications-part-1

https://gradientflow.com/best-practices-in-retrieval-augmented-generation/

https://stackoverflow.blog/2023/10/18/retrieval-augmented-generation-keeping-llms-relevant-and-current/

https://www.rungalileo.io/blog/optimizing-llm-performance-rag-vs-finetune-vs-both

I use the above blog as reference and discuss on when to use RAG or Fine-Tuning or Both for improving performance of LLMs

In this video I explain how you can create a medical chatbot/converse with medical data using LlamaIndex and Mistral 7B LLM. I also explain how you can use custom embeddings ( sentence transformers )

Colab Notebook:

https://colab.research.google.com/drive/12k7RVPmGpPfFNXY6fGvqtMoAVOtjz44o?usp=sharing

Relevant Links:

https://huggingface.co/TheBloke/Mistral-7B-Instruct-v0.1-GGUF

https://mistral.ai/news/announcing-mistral-7b/

https://huggingface.co/thenlper/gte-large

https://gpt-index.readthedocs.io/en/stable/core_modules/model_modules/llms/modules.html#llama-cpp

https://python.langchain.com/docs/integrations/llms/llamacpp

https://gpt-index.readthedocs.io/en/stable/index.html

https://www.versusarthritis.org/media/24901/fibromyalgia-information-booklet-july2021.pdf

The Mistral-7B-Instruct-v0.1 Large Language Model (LLM) is a instruct fine-tuned version of the Mistral-7B-v0.1 generative text model using a variety of publicly available conversation datasets.

LlamaIndex is a "data framework" to help you build LLM apps. It provides the following tools:

· Offers data connectors to ingest your existing data sources and data formats (APIs, PDFs, docs, SQL, etc.)

· Provides ways to structure your data (indices, graphs) so that this data can be easily used with LLMs.

· Provides an advanced retrieval/query interface over your data: Feed in any LLM input prompt, get back retrieved context and knowledge-augmented output.

· Allows easy integrations with your outer application framework (e.g. with LangChain, Flask, Docker, ChatGPT, anything else).

LlamaIndex provides tools for both beginner users and advanced users. Our high-level API allows beginner users to use LlamaIndex to ingest and query their data in 5 lines of code. Our lower-level APIs allow advanced users to customize and extend any module (data connectors, indices, retrievers, query engines, reranking modules), to fit their needs.

In this video I make of use of Unsloth Colab notebook to finetune Google Gemma 3 4B LLM on Medical dataset Colab notebook: https://colab.research.google.com/drive/1gvmrL0vnUq7EVZ-5twVvqwISnqZifiWO?usp=sharing Unsloth colab notebook: https://colab.research.google.com/github/unslothai/notebooks/blob/main/nb/Gemma3_(4B).ipynb https://unsloth.ai/blog/gemma3 https://huggingface.co/datasets/lavita/ChatDoctor-HealthCareMagic-100k https://huggingface.co/docs/peft/main/en/conceptual_guides/lora

In this video with the help of a Colab notebook from Unsloth AI, I explain how you can fine tune the recently released Qwen-3 family of LLMs on a mix of reasoning and non-reasoning datasets Colab notebook:https://colab.research.google.com/gist/rsreetech/3cfecac2425b28894a5fbe827eae2129/qwen3_-14b-reasoning-conversational.ipynb

In intelligence and artificial intelligence, an intelligent agent (IA) is an agent that perceives its environment, takes actions autonomously in order to achieve goals, and may improve its performance with learning or acquiring knowledge. A Generative AI agent can be defined as an application that attempts to achieve a goal by observing the world and acting upon it using the tools that it has at its disposal. Agents are autonomous and can act independently of human intervention, especially when provided with proper goals or objectives they are meant to achieve. Agents can also be proactive in their approach to reaching their goals. Even in the absence of explicit instruction sets from a human, an agent can reason about what it should do next to achieve its ultimate goal.

Relevant Links:

https://www.kaggle.com/whitepaper-agents

https://aws.amazon.com/what-is/ai-agents/

https://news.microsoft.com/source/features/ai/ai-agents-what-they-are-and-how-theyll-change-the-way-we-work/

https://en.wikipedia.org/wiki/Intelligent_agent