# Origins AI

> Bringing Trust and Speed back in Software Services


## Posts

- [Anon](https://originshq.com/our-works/anon/)
- [Equi](https://originshq.com/our-works/equi/): From Spreadsheet Chaos to Live Risk Intelligence — Equi | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam [&hellip;]
- [Immigr8](https://originshq.com/our-works/immigr8/): From Manual Visa Casework to Automation-Driven Immigration Processing — Immigr8 | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming [&hellip;]
- [Refold](https://originshq.com/our-works/refold/): From Fragmented DevOps to a Cloud-Agnostic Deployment Foundation — Refold AI | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam [&hellip;]
- [Zoca](https://originshq.com/our-works/zoca/): From Manual Website Setup to AI-Generated Sites at Scale — Zoca AI | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes [&hellip;]
- [Nubon](https://originshq.com/our-works/nubon/): From Off-the-Shelf Ecommerce to a Custom Commerce Platform — Nubon | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming [&hellip;]
- [Understanding DeepSeek R1 Training: A New Era in Reasoning AI](https://originshq.com/blog/understanding-deepseek-r1-training/): DeepSeek R1 is a remarkable step forward in large language model training, known for its unique multi-stage reinforcement learning process and ability to create reasoning-focused AI. This blog dives into the training methodology of DeepSeek R1, demystifying its capabilities and showcasing its relevance in advancing AI technologies. What is DeepSeek R1? DeepSeek R1 represents a significant evolution in reasoning capabilities for large language models (LLMs). Developed by DeepSeek-AI, this model stands out for its multi-stage training pipeline, which involves cold-start data, reinforcement learning (RL), and fine-tuning to optimize its performance on complex reasoning tasks. DeepSeek R1 has been benchmarked to [&hellip;]
- [How to Fine-Tune Language Models First Principles to Scalable Performance](https://originshq.com/blog/how-to-fine-tune-language-models-beginner-to-expert/): In this article, well explore the process of fine-tuning language models for text classification. Well do so in three levels: first, by manually adding a classification head in PyTorch* and training the model so you can see the full process; second, by using the Hugging Face* Transformers library to streamline the process; and third, by leveraging PyTorch Lightning* and accelerators to optimize training performance. By the end of this guide, you&#8217;ll have a well-rounded understanding of the fine-tuning workflow. Introduction The concept of using fine-tuning in Natural Language Processing (NLP) was borrowed from Computer Vision (CV). The CV models were [&hellip;]
- [Going Beyond the 1000-Layer Convolution Network](https://originshq.com/blog/going-beyond-the-1000-layer-convolution-network/): Mean gradient for 1st layer in all experiments Introduction One of the largest Convolutional Networks, ConvNext-XXLarge [1] from OpenCLIP[2], boasts approximately 850 million parameters and 120 layers (counting all convolutional and linear layers). This is a dramatic increase compared to the 8 layers of AlexNet[3] but still fewer than the 1001-layer experiment introduced in the PreResNet[4] paper. Interestingly, about a decade ago, training networks with more than 100 layers was considered nearly impossible due to the vanishing gradient problem. However, advancements such as improved activation functions, normalization layers, and skip connections have significantly mitigated this issue or so it seems. [&hellip;]
- [Logistic Regression A Simple Guide to Intuition and Implementation in Python](https://originshq.com/blog/logistic-regression-a-simple-guide-to-intuition-and-implementation-in-python/): When it comes to solving classification problems, logistic regression is often the first algorithm that comes to our mind. The theoretical concepts of logistic regression are essential for understanding more advanced concepts in deep learning. Lets get Started Introduction: Logistic regression is a fundamental classification algorithm used to predict the probability of categorical dependent variable. The idea of logistic regression is to find the relationship between independent variables and the probability of dependent variables. Simply put, it is a classification algorithm used when the response variable is categorical typically binary (e.g. 0 or 1). A Simple Example Suppose you have [&hellip;]
- [Building a Local Committee-of-Expert (CoE) RAG Application for Document Discovery](https://originshq.com/blog/building-a-local-committee-of-expert-coe/): In todays fast-paced world, where access to timely and accurate information can be a critical differentiator, organizations across various sectors constantly seek innovative solutions to stay ahead of the competition. This is particularly true in the insurance and reinsurance industry where the underwriting expenses have grown significantly in the last decade. Rising costs, salaries, and regulatory compliance requirements drive this consistent increase in underwriting expenses. U.S. Property and Casualty Insurance Industry Underwriting Expenses (Source: WWW.NAIC.ORG) Insurance companies spend considerable time and resources in parsing out the information in the policy and claims forms. The extracted information from these documents underpins [&hellip;]
- [Building Large Action Models Insights from Microsoft](https://originshq.com/blog/building-large-action-models-insights-from-microsoft/): Action execution is one of the key building blocks of agentic workflows. One of the most interesting debates in that are is whether actions are executed by the model itself or by an external coordination layer. The supporters of the former hypothesis have lined up behind a theory known as large action models(LAMs) with projects like Gorilla or Rabbit r1 as key pioneers. However, there are still only a few practical examples of LAM frameworks. Recently, Microsoft Research published one of the most complete papers in this area outlining a complete framework for LAM models. Microsofts core idea is to [&hellip;]
- [Machine Learning (ML) vs. Artificial Intelligence (AI)  Crucial Differences](https://originshq.com/blog/machine-learning-vs-artificial-intelligence/): Helping Scale AI &amp; Technology Startups to Enterprises Recently, a report was released regarding the misuse of companies claiming to use artificial intelligence [29] [30] on their products and services. According to the Verge [29], 40% of European startups claiming to use AI dont use the technology. Last year, TechTalks, also stumbled upon such misuse by companies claiming to use machine learning and advanced artificial intelligence to gather and examine thousands of users data to enhance user experience in their products and services [2] [33]. Unfortunately, theres still much confusion within the public and the media regarding what genuinely is [&hellip;]
- [Showcasing Soaring Wildfire Counts With Streamlit and Python: A Powerful Approach](https://originshq.com/blog/showcasing-soaring-wildfire-counts-with-streamlit-and-python/): Python ** Streamlit** is terrific for creating interactive maps from a GIS dataset. Interactive maps that allow input from your audience can be used for deeper analysis and storytelling. Python Streamlit is the right tool for the job. It can be used alongside the pandas for easy data frame creation and manipulation. Let’s test this out with a deep and detailed dataset on a very prescient issue — the seeming escalation of wildfires. There is a terrific public wildfires dataset available on the site managed by Natural Resources Canada. With this detailed dataset let’s take a modular approach to our [&hellip;]
- [How to Pick Between Data Science, Data Analytics, Data Engineering, ML Engineering, and SW Engineering in 2025](https://originshq.com/blog/how-to-pick-between-data-science-and-other-data-roles/): In 2025 If you’re hoping to break into tech or pivot into a new job family, figuring out which career path is right for you isn’t easy — especially when the job titles sound so similar and the roles have a good amount of overlap. Data analytics, data engineering, data science, machine learning engineering, software engineering — while there are definitely similarities, the day-to-day work and the type of person who thrives in each role couldn’t be more different. And if you choose wrong, you could end up spending tons of time and money building skills you’ll never use, or [&hellip;]
- [Top AI/LLM learning resource in 2025](https://originshq.com/blog/top-ai-llm-learning-resource-in-2025/): The Blog is organized into three main segments: 📝 Notebooks Below is a collection of notebooks and articles dedicated to LLMs. Tools Notebook Name Description Notebook 🧐 LLM AutoEval Evaluate your LLMs automatically using RunPod. Notebook 🥱 LazyMergekit Merge models effortlessly using MergeKit with a single click. Notebook 🦎 LazyAxolotl Fine-tune models in the cloud with Axolotl in just one click. Notebook ⚡ AutoQuant Quantize LLMs into GGUF, GPTQ, EXL2, AWQ, and HQQ formats in one click. Notebook 🌳 Model Family Tree Visualize the lineage of merged models. Notebook 🚀 ZeroSpace Instantly create a Gradio chat interface using a free [&hellip;]
- [Improve ChatGPT with Knowledge Graphs](https://originshq.com/blog/improve-chatgpt-with-knowledge-graphs/): ChatGPT has shown impressive capabilities in processing and generating human-like text. However, it is not without its imperfections. A primary concern is the model’s propensity to produce either inaccurate or obsolete answers , often called “hallucinations.” The New York Times recently highlighted this issue in their article, “Here’s What Happens When Your Lawyer Uses ChatGPT.” It presents a lawsuit where a lawyer leaned heavily on ChatGPT to assist in preparing a court filing for a client suing an airline. The model generated fictional court decisions to back its arguments, which didn’t go unnoticed. This incident underscores the need for solutions [&hellip;]
- [Uncensor any LLM with abliteration](https://originshq.com/blog/uncensor-any-llm-with-abliteration/): The third generation of Llama models provided fine-tunes (Instruct) versions that excel in understanding and following instructions. However, these models are heavily censored, designed to refuse requests seen as harmful with responses such as “As an AI assistant, I cannot help you.” While this safety feature is crucial for preventing misuse, it limits the model’s flexibility and responsiveness. In this article, we will explore a technique called “abliteration” that can uncensor any LLM without retraining. This technique effectively removes the model’s built-in refusal mechanism, allowing it to respond to all types of prompts. The code is available on Google Colab [&hellip;]
- [Create Mixtures of Experts with MergeKit](https://originshq.com/blog/create-mixtures-of-experts-with-mergekit/): Thanks to the release of Mixtral, the Mixture of Experts (MoE) architecture has become popular in recent months. This architecture offers an interesting tradeoff: higher performance at the cost of increased VRAM usage. While Mixtral and other MoE architectures are pre-trained from scratch, another method of creating MoE has recently appeared. Thanks to Arcee’s MergeKit library, we now have a new way of creating MoEs by ensembling several pre-trained models. These are often referred to as frankenMoEs or MoErges to distinguish them from the pre-trained MoEs. In this article, we will detail how the MoE architecture works and how frankenMoEs [&hellip;]
- [Merge Large Language Models with MergeKit](https://originshq.com/blog/merge-large-language-models-with-mergekit/): Model merging is a technique that combines two or more LLMs into a single model. It’s a relatively new and experimental method to create new models for cheap (no GPU required). Model merging works surprisingly well and produced many state-of-the-art models on the Open LLM Leaderboard. In this tutorial, we will implement it using the mergekit library. More specifically, we will review four merge methods and provide examples of configurations. Then, we will use mergekit to create our own model, Marcoro14-7B-slerp, which became the best-performing model on the Open LLM Leaderboard (02/01/24). The code is available on GitHub and Google [&hellip;]
- [ExLlamaV2 The Fastest Library to Run LLMs](https://originshq.com/blog/exllamav2-the-fastest-library-to-run-llms/): Quantizing Large Language Models (LLMs) is the most popular approach to reduce the size of these models and speed up inference. Among these techniques, GPTQ delivers amazing performance on GPUs. Compared to unquantized models, this method uses almost 3 times less VRAM while providing a similar level of accuracy and faster generation. It became so popular that it has recently been directly integrated into the transformers library. ExLlamaV2 is a library designed to squeeze even more performance out of GPTQ. Thanks to new kernels, it’s optimized for (blazingly) fast inference. It also introduces a new quantization format, EXL2, which brings [&hellip;]
- [Decoding Strategies in Large Language Models](https://originshq.com/blog/decoding-strategies-in-large-language-models/): In the fascinating world of large language models (LLMs), much attention is given to model architectures, data processing, and optimization. However, decoding strategies like beam search, which play a crucial role in text generation, are often overlooked. In this article, we will explore how LLMs generate text by delving into the mechanics of greedy search and beam search, as well as sampling techniques with top-k and nucleus sampling. By the conclusion of this article, you’ll not only understand these decoding strategies thoroughly but also be familiar with how to handle important hyperparameters like temperature, , , and . The code [&hellip;]
- [Quantize Llama models with GGUF and llama.cpp](https://originshq.com/blog/quantize-llama-models-with-gguf-and-llama-cpp/): Due to the massive size of Large Language Models (LLMs), quantization has become an essential technique to run them efficiently. By reducing the precision of their weights, you can save memory and speed up inference while preserving most of the model’s performance. Recently, 8-bit and 4-bit quantization unlocked the possibility of running LLMs on consumer hardware. Coupled with the release of Llama models and parameter-efficient techniques to fine-tune them (LoRA, QLoRA), this created a rich ecosystem of local LLMs that are now competing with OpenAI’s GPT-3.5 and GPT-4. Currently, there are three main quantization techniques: NF4, GPTQ, and GGML. NF4 [&hellip;]
- [4-bit LLM Quantization with GPTQ](https://originshq.com/blog/4-bit-llm-quantization-with-gptq/): Recent advancements in weight quantization allow us to run massive large language models on consumer hardware, like a LLaMA-30B model on an RTX 3090 GPU. This is possible thanks to novel 4-bit quantization techniques with minimal performance degradation, like GPTQ, GGML, and NF4. 🧠 Optimal Brain Quantization For every layer \( \ell \) in the network, we want to find a quantized version \( \widehat{\mathbf{W}}_\ell \) of the original weights \( \mathbf{W}_\ell \). This is called the **layer-wise compression problem**. More specifically, to minimize performance degradation, we want the outputs \( \widehat{\mathbf{W}}_\ell \mathbf{X}_\ell \) of these new weights to be [&hellip;]
- [Introduction to Weight Quantization](https://originshq.com/blog/introduction-to-weight-quantization/): Large Language Models (LLMs) are known for their extensive computational requirements. Typically, the size of a model is calculated by multiplying the number of parameters ( size ) by the precision of these values ( data type ). However, to save memory, weights can be stored using lower-precision data types through a process known as quantization. We distinguish two main families of weight quantization techniques in the literature: In this article, we focus on PTQ to reduce the precision of our parameters. To get a good intuition, we will apply both naïve and more sophisticated techniques to a toy example [&hellip;]
- [Step-by-step guide to supervised fine-tune Llama 2 in Google Colab.](https://originshq.com/blog/step-by-step-guide-to-supervised-fine-tune-llama-2-in-google-colab/): With the release of LLaMA v1, we saw a Cambrian explosion of fine-tuned models, including Alpaca, Vicuna, WizardLM, among others. This trend encouraged different businesses to launch their own base models with licenses suitable for commercial use, such as OpenLLaMA, Falcon, XGen, etc. The release of Llama 2 now combines the best elements from both sides: it offers a highly efficient base model along with a more permissive license. During the first half of 2023, the software landscape was significantly shaped by the widespread use of APIs (like OpenAI API) to create infrastructures based on Large Language Models (LLMs). Libraries [&hellip;]
- [End-to-end guide to the state-of-the-art tool for fine-tuning.](https://originshq.com/blog/end-to-end-guide-to-the-state-of-the-art-tool-for-fine-tuning/): The growing interest in Large Language Models (LLMs) has led to a surge in tools and wrappers designed to streamline their training process. Popular options include FastChat from LMSYS (used to train Vicuna) and Hugging Face’s transformers/trl libraries (used in my previous article). In addition, each big LLM project, like WizardLM, tends to have its own training script, inspired by the original Alpaca implementation. In this article, we will use Axolotl, a tool created by the OpenAccess AI Collective. We will use it to fine-tune a Code Llama 7b model on an evol-instruct dataset comprised of 1,000 samples of Python [&hellip;]
- [Boost the performance of supervised fine-tuned models with DPO.](https://originshq.com/blog/boost-the-performance-of-supervised-fine-tuned-models-with-dpo/): Pre-trained Large Language Models (LLMs) can only perform next-token prediction, making them unable to answer questions. This is why these base models are then fine-tuned on pairs of instructions and answers to act as helpful assistants. However, this process can still be flawed: fine-tuned LLMs can be biased, toxic, harmful, etc. This is where Reinforcement Learning from Human Feedback (RLHF) comes into play. RLHF provides different answers to the LLM, which are ranked according to a desired behavior (helpfulness, toxicity, etc.). The model learns to output the best answer among these candidates, hence mimicking the behavior we want to instill. [&hellip;]
- [Fine-tune Llama 3 with ORPO](https://originshq.com/blog/fine-tune-llama-3-with-orpo/): ORPO is a new exciting fine-tuning technique that combines the traditional supervised fine-tuning and preference alignment stages into a single process. This reduces the computational resources and time required for training. Moreover, empirical results demonstrate that ORPO outperforms other alignment methods on various model sizes and benchmarks. In this article, we will fine-tune the new Llama 3 8B model using ORPO with the TRL library. The code is available on Google Colab and in the LLM Course on GitHub. ORPO Instruction tuning and preference alignment are essential techniques for adapting Large Language Models (LLMs) to specific tasks. Traditionally, this involves [&hellip;]
- [Fine-tune Llama 3.1 Ultra-Efficiently with Unsloth](https://originshq.com/blog/fine-tune-llama-3-1-ultra-efficiently-with-unsloth/): A beginner&#8217;s guide to state-of-the-art supervised fine-tuning The recent release of Llama 3.1 offers models with an incredible level of performance, closing the gap between closed-source and open-weight models. Instead of using frozen, general-purpose LLMs like GPT-4o and Claude 3.5, you can fine-tune Llama 3.1 for your specific use cases to achieve better performance and customizability at a lower cost. In this article, we will provide a comprehensive overview of supervised fine-tuning. We will compare it to prompt engineering to understand when it makes sense to use it, detail the main techniques with their pros and cons, and introduce major [&hellip;]
- [Transformer squared: Self-Adaptive LLMs](https://originshq.com/blog/transformer%c2%b2-self-adaptive-llms/): Intro Adaptation is one of the most remarkable phenomena in nature. From the way an octopus can change their skin color to blend into its surroundings, to how the human brain rewires itself after an injury, allowing individuals to recover lost functions and adapt to new ways of thinking or moving. Living organisms exhibit adaptability that allows life to flourish in diverse and ever-changing environments. In the field of AI, the concept of adaptation holds a similar allure. Imagine a machine learning system that could adjust its own weights dynamically to thrive in unfamiliar settings, essentially illustrating a system that [&hellip;]
- [Migrating YesMadam from Android/IOS to React Native](https://originshq.com/our-works/migrating-yesmadam-from-android-ios-to-react-native/): OriginsAI Powers Yes Madam About Us Blogs Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage Helping FrontPage Scale &#8599; NuCash Building Platform at Nucash &#8599; RagaAI Building an Ai Testing Platform &#8599; Services Digital Transformation Seamless Digitization For Your Business’s Success &#8599; Data Services Building Intelligent Solutions For The Future &#8599; Artificial Intelligence Driving Innovation With Intelligent And Scalable Solutions &#8599; Cloud Redefine Scalability With Unparalleled Flexibility And Performance &#8599; DevOps Unleashing Agility For Unprecedented Growth [&hellip;]
- [Scaling AI in the Classroom: A Khan Academy Case Study with Khanmigo](https://originshq.com/blog/scaling-ai-at-khan-academy-case-study/): Khan Academy&#8217;s mission is to provide free, world-class education to everyone, everywhere. This blog post explores Khanmigo, our AI tutor and teacher assistant, and how it helps us achieve this mission at scale. We&#8217;ll delve into the problems we&#8217;re tackling, our solutions, technical challenges, and future plans. The Challenge: Bridging the Widening Learning Gap The pandemic exacerbated existing learning gaps. Data reveals that while many third-graders perform at grade level, a significant number fall behind by eighth grade. Khan Academy believes AI can help close these gaps by providing personalized tutoring at scale. Khanmigo, our AI tutor for learners and [&hellip;]
- [From DevOps to AIOps: Building a Scalable AI Platform](https://originshq.com/blog/from-devops-to-aiops/): The evolution of software development has seen a constant influx of new paradigms. From Agile to DevOps, each shift has aimed to optimize processes and enhance collaboration. Today, Artificial Intelligence (AI) stands as the next transformative force. This blog post explores the journey from DevOps to AIOps, emphasizing the importance of a robust AI platform for scalable AI adoption. It delves into the key components of such a platform, the necessary enablement practices, effective governance, and the changing role of engineers in this AI-driven world. The Rise of the AI Engineer and the Need for a Platform The term &#8220;AI [&hellip;]
- [From Structured Data to the Unstructured Data Revolution: Navigating the LLM Landscape](https://originshq.com/blog/from-structured-data-to-the-unstructured-llm-landscape/): The AI field has evolved dramatically. For over a decade, structured data fueled AI value. Supervised and deep learning excelled at labeling, but this approach overlooks a crucial reality: most organizational data is unstructured. This blog post explores the shift towards leveraging unstructured data with Large Language Models (LLMs), addresses the hype around Artificial General Intelligence (AGI), and examines the challenges and technical debt associated with LLM deployment. The Untapped Potential of Unstructured Data Estimates suggest over 80% of organizational data is unstructured. A significant 71% of organizations struggle to manage and secure this data. Previously, building automated systems to [&hellip;]
- [Navigating the Real-World Risks of LLM Deployment in the Enterprise](https://originshq.com/blog/risks-of-llm-deployment/): The promise of AI co-pilots augmenting our work is captivating. However, the reality of enterprise AI adoption often presents a different picture. This post explores the practical challenges and solutions for deploying secure and accurate large language model (LLM) applications in real-world business settings. The LLM Landscape and Its Challenges Open-access large language models are increasingly central to enterprise AI strategies. While powerful, these models introduce unique risks requiring careful consideration. We will explore these risks based on real-world experiences and offer practical mitigation strategies. Checklist of LLM Deployment Risks Here&#8217;s a checklist of common challenges encountered when deploying LLMs: [&hellip;]
- [Measuring the ROI of AI Coding Tools: Beyond the Hype](https://originshq.com/blog/ai-coding-tools-beyond-the-hype/): The buzz around AI coding tools is undeniable. Promising increased productivity and reduced development costs, these tools are quickly becoming a staple in the software development landscape. But amidst the excitement, a crucial question lingers: how do we actually measure their real impact on our bottom line? This is a challenge many engineering leaders grapple with, especially when justifying the cost of these new tools to stakeholders. This blog post delves into the complexities of measuring AI ROI in software development, drawing insights from Bang Liu, CTO of Sourcegraph, and his experiences working with a diverse range of organizations. The [&hellip;]
- [The Democratization of AI: From Demo to Production](https://originshq.com/blog/from-demo-to-production/): The world of Artificial Intelligence is rapidly evolving. AI is no longer confined to academic research or niche tech companies. It&#8217;s permeating every industry, from healthcare and agriculture to manufacturing and beyond. This democratization is largely thanks to large language models (LLMs) and conversational interfaces, making AI accessible to a broader audience. But a significant challenge remains: bridging the gap between impressive demos and robust production applications. This article explores the nuances of productionizing AI, drawing insights from real-world experiences and industry trends. The AI Production Paradox: Easy to Demo, Hard to Deploy Building a compelling AI demo is surprisingly [&hellip;]
- [Decoding LLM Inference: A Deep Dive into Workloads, Optimization, and Cost Control](https://originshq.com/blog/demystifying-llm-inference/): Large language models (LLMs) have revolutionized how we interact with technology. But deploying these powerful models for inference can be complex and costly. This blog post delves into the intricacies of LLM inference, providing a clear understanding of the workload, key performance metrics, and practical optimization strategies. Understanding the LLM Inference Workload At its core, LLM inference involves sending a prompt to the GPU and generating tokens one at a time. The GPU retains the entire prompt and every generated token in its memory. This mechanism allows the LLM to maintain context and generate coherent responses. Let&#8217;s break down the [&hellip;]
- [Power of Amazon S3: Transforming Storage into a Data Lakehouse and Avoiding Costly Pitfalls](https://originshq.com/blog/power-of-aws-s3-costly-pitfalls/): Amazon S3 is evolving beyond traditional storage, emerging as a robust data lakehouse solution with the introduction of S3 Iceberg Tables and S3 Metadata. In this comprehensive guide, we explore how these innovations leverage the Apache Iceberg open table format to enhance data management, optimize query performance, and integrate seamlessly with a variety of data processing engines. Discover the advanced features, performance benefits, and strategic advantages that position S3 as your new database of choice.  But with great power comes great responsibility. In Appendix A, we delve into a cautionary tale by Maciej Pocwierz, illustrating how misconfigurations in S3 buckets can lead to unexpectedly exorbitant AWS bills. Learn from real-world experiences to implement best practices that safeguard your AWS resources, prevent unauthorized access, and maintain control over your cloud expenditures. Whether you're harnessing the latest S3 advancements or fortifying your storage setup, this guide provides essential insights to maximize the potential of Amazon S3 while mitigating risks.
- [Building an AI Testing Platform](https://originshq.com/our-works/building-an-ai-testing-platform/): From Rigid Testing Bottleneck to Scalable AI Evaluation Platform — RagaAI | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam [&hellip;]
- [Samsung IoT Development and User Verification](https://originshq.com/our-works/samsung-iot-development-and-user-verification/): Software Development-Samsung About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage Helping FrontPage Scale &#8599; [&hellip;]
- [Cost Reduction at Amazon](https://originshq.com/our-works/cost-reduction-at-amazon/): Enhancing Efficiency-Amazon About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage Helping FrontPage Scale &#8599; [&hellip;]
- [Transforming YesMadam into a Tech Company](https://originshq.com/our-works/transforming-yesmadam-into-a-tech-company/): Origins AI Powers Yes Madam About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage Helping [&hellip;]
- [Building Infra at PlayStation](https://originshq.com/our-works/building-infra-at-playstation/): Software Development-Sony About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage Helping FrontPage Scale &#8599; [&hellip;]
- [Helping FrontPage Scale](https://originshq.com/our-works/helping-frontpage-scale/): Origins AI Powers FrontPage About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage Helping FrontPage [&hellip;]
- [Building Platform at Nucash](https://originshq.com/our-works/building-platform-at-nucash/): Origins AI Powers NuCash About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage Helping FrontPage [&hellip;]
- [Kolmogorov-Arnold Networks: A Powerful Alternative to MLPs](https://originshq.com/blog/kolmogorov-arnold-networks-a-powerful-alternative-to-mlps/): MLPs come with limitations, particularly in terms of accuracy and interpretability. This is where Kolmogorov-Arnold Networks (KANs) step in, offering a compelling alternative inspired by a powerful mathematical theorem.
- [Decoupled Weight Decay Regularization: Bye Bye Adam Optimizer](https://originshq.com/blog/decoupled-weight-decay-regularization-bye-bye-adam-optimizer/): Boost your neural network's performance with AdamW! Learn how decoupled weight decay can significantly improve Adam optimizer's generalization ability, leading to better results and easier hyperparameter tuning.
- [ReALM: Reference Resolution with Language Modelling](https://originshq.com/blog/realm-reference-resolution-with-language-modelling/): This blog explores ReALM, a groundbreaking approach by Apple researchers that utilizes language models to resolve references to both conversational and on-screen entities. ReALM outperforms existing systems and achieves accuracy comparable to OpenAI's GPT-4, paving the way for more natural and intuitive interactions with voice assistants and conversational AI. Click to learn how ReALM is revolutionizing reference resolution and shaping the future of human-machine interaction.
- [Demystifying LLMs: A Deep Dive into Large Language Models](https://originshq.com/blog/demystifying-llms-a-deep-dive-into-large-language-models/): This blog post will delve into the intricacies of LLMs, exploring their inner workings, capabilities, future directions, and potential security concerns.
- [Building a LLM in 2024: A Detailed Guide](https://originshq.com/blog/building-a-llm-in-2024-a-detailed-guide/): This guide delves into the process of building an LLM from scratch, focusing on the often-overlooked aspects of training and data preparation. We'll also touch on fine-tuning, inference, and the importance of sharing your work with the community.
- [LISA: A Simple But Powerful Way to Fine-Tune LLM Efficiently](https://originshq.com/blog/lisa-a-simple-but-powerful-way-to-fine-tune-llm-efficiently/): LISA introduces a surprisingly simple yet effective strategy for fine-tuning LLMs. It builds upon a key observation about LoRA: the weight norms across different layers exhibit an uncommon skewness. The bottom and top layers tend to dominate the updates, while the middle layers contribute minimally.
- [QMoE: Bringing Trillion-Parameter Models to Commodity Hardware](https://originshq.com/blog/qmoe-bringing-trillion-parameter-models-to-commodity-hardware/): This blog post delves into QMoE, a novel compression and execution framework that tackles the memory bottleneck of massive MoEs. QMoE introduces a scalable algorithm that compresses trillion-parameter MoEs to less than 1 bit per parameter, utilizing a custom format and bespoke GPU decoding kernels for efficient end-to-end compressed inference.
- [AnimateDiff: Paper Explained](https://originshq.com/blog/animatediff-paper-explained/): Introducing AnimateDiff, a groundbreaking framework that empowers you to animate your personalized T2I models without the need for complex, model-specific tuning. This means you can now breathe life into your unique creations and watch them come alive in smooth, visually-appealing animations.
- [Jamba : A hybrid model (GPT + Mamba) by AI 21 Labs](https://originshq.com/blog/jamba-a-hybrid-model-gpt-mamba-by-ai-21-labs/): Jamba boasts a 256K context window, allowing it to consider a vast amount of preceding information when processing a task. This extended context window is particularly beneficial for tasks requiring a deep understanding of a conversation or passage.
- [DBRX: A New State-of-the-Art Open LLM by Databricks](https://originshq.com/blog/introducing-dbrx-a-new-state-of-the-art-open-llm/): DBRX utilizes a transformer-based decoder-only architecture with a fine-grained Mixture-of-Experts (MoE) design. This means it uses a large number of smaller expert models to process different parts of the input, rather than relying on a single massive model.
- [Fine-tune an Instruct model over raw text data](https://originshq.com/blog/fine-tune-an-instruct-model-over-raw-text-data/): This experiment seeks to discover a lighter approach that navigates between the constraints of a 128K context window and the complexities of a model fine-tuned on billions of tokens, perhaps more in the realm of tens of millions of tokens. For a smaller-scale test, I’ll fine-tune Mistral’s 7B Instruct v0.2 model on The Guardian’s manage-frontend repository (the dataset being 1.6 million tokens).
- [Crew AI Tutorial](https://originshq.com/blog/crew-ai-tutorial/): In the realm of artificial intelligence, the adoption of multi-agent systems (MAS) via crew ai represents a paradigm shift towards more dynamic and complex problem-solving capabilities. This blog dives into the essence of Multi Agent Systems, highlighting the necessity for such systems in today’s technological landscape and exploring the CrewAI framework as a possible solution.
- [Fine Tuning LLMs on a Single Consumer Graphic Card](https://originshq.com/blog/fine-tuning-llms-on-a-single-consumer-graphic-card/): Unlock the secrets of fine-tuning large language models (LLMs) on a single consumer graphics card. Discover how to leverage your GPU for enhanced AI model performance without breaking the bank. Ideal for AI enthusiasts and developers seeking practical insights.
- [LangChain Cheatsheet — All Secrets on a Single Page](https://originshq.com/blog/langchain-cheatsheet-all-secrets-on-a-single-page/): Unlock the full potential of LangChain with our ultimate cheatsheet! Discover all the secrets and tips you need, condensed into a single, easy-to-navigate page. Save time and boost your efficiency today!
- [TaCo: Enhancing Cross-Lingual Transfer for Low-Resource Languages in LLM](https://originshq.com/blog/taco-enhancing-cross-lingual-transfer-for-low-resource-languages-in-llm/): TaCo, short for Translation-Assisted Cross-Linguality, utilizes a clever combination of translation and chain-of-thought processes to efficiently train LLMs on new languages. This blog post delves into the details of TaCo, exploring its approach, datasets, training process, and evaluation results.
- [RAFT: Finetuning LLM for Domain-Specific Knowledge](https://originshq.com/blog/raft-finetuning-llm-for-domain-specific-knowledge/): This blog post dives into RAFT (Retrieval-Augmented Fine-Tuning), a novel training recipe that enhances the ability of LLMs to perform well in domain-specific RAG tasks. We'll explore the challenges of adapting LLMs to specialized domains, delve into the details of RAFT, and analyze its effectiveness through evaluation results and qualitative examples.
- [Open Source Techniques Unlock Hidden Potential in LLM](https://originshq.com/blog/open-source-techniques-unlock-hidden-potential-in-llm/): Matt Shumer, CEO of Hyperwrite AI, introduced the "Claude Opus to Haiku" technique, which allows users to achieve Claude 3 Opus-level quality at a fraction of the cost and latency. This open-source method involves providing Claude 3 Opus with a task description and a single input-output example.
- [MoonDream: A Tiny Vision Language Model for Edge Devices](https://originshq.com/blog/moondream-a-tiny-vision-language-model-for-edge-devices/): MoonDream is a small, efficient vision language model (VLM) that can operate with less than 6GB of memory. This makes it ideal for use on edge devices in a variety of applications, such as:
- [Run LLM Locally with Ollama: A Hands-on Guide](https://originshq.com/blog/run-llm-locally-with-ollama-a-hands-on-guide/): The rise of generative AI has opened doors to exciting possibilities, with large language models (LLMs) at the forefront. However, experimenting with different LLMs can be challenging, especially when dealing with cloud-based solutions. This is where Ollama comes in, offering a powerful and convenient way to run open-source LLMs locally on your own system.
- [Build a Video Analysis Engine with Moondream : Tutorial](https://originshq.com/blog/build-a-video-analysis-engine-with-moondream-tutorial/): The ability to understand and interpret visual information is a crucial aspect of artificial intelligence. Moondream, a powerful multi-modal model, offers a promising solution for building video understanding engines. This
- [NVIDIA Unveils Blackwell Platform and a Slew of AI Innovations](https://originshq.com/blog/nvidia-unveils-blackwell-platform-and-a-slew-of-ai-innovations/): NVIDIA's GTC 2023 keynote was a whirlwind of exciting announcements, showcasing the company's commitment to pushing the boundaries of AI and robotics. CEO Jensen Huang unveiled groundbreaking technologies, including the new Blackwell platform, advancements in robotics with Project GROOT, and collaborations with industry giants like Google, Amazon, and Microsoft.
- [2024: The Year of Humanoid Robots?](https://originshq.com/blog/2024-the-year-of-humanoid-robots/): The year 2024 is shaping up to be a landmark year for robotics, particularly in the field of humanoid robots. Several major players in the industry, including OpenAI, NVIDIA, Boston Dynamics, and Anduril, have recently announced exciting advancements that point towards a future where robots play a much larger role in our lives.
- [Unearthing Hidden Gems: 8 Under the Radar AI Tools](https://originshq.com/blog/unearthing-hidden-gems-8-under-the-radar-ai-tools/): The AI landscape is constantly evolving, with new tools and advancements emerging daily. While major players like OpenAI, Google, and Microsoft often dominate the headlines, a plethora of smaller, innovative AI tools are bubbling under the surface, waiting to be discovered.
- [Unveiling Genie: Google's Revolutionary Image-to-Game AI](https://originshq.com/blog/genie-googles-image-to-game-ai/): Google&#8217;s latest innovation, Genie, stands at the forefront of AI technology, transforming images, sketches, and photographs into interactive video game worlds. This groundbreaking AI model not only revolutionizes content creation but also opens a new chapter in interactive virtual world development, making game creation as effortless as imagining it. Transforming Imagination into Reality Genie&#8217;s magic lies in its ability to bring static images to life. By turning drawings or photos into playable video game levels, it offers creators, gamers, and educators a new realm of possibilities. This AI doesn&#8217;t just push boundaries; it redefines them, showcasing a future where creating [&hellip;]
- [Ideogram AI 1.0: A New Frontier in Image Generation](https://originshq.com/blog/ideogram-ai-1-0/): The world of artificial intelligence never ceases to amaze, with innovations emerging at a breathtaking pace. Among the latest entrants in the AI space, Ideogram AI&#8217;s release of Ideogram 1.0 has stirred the waters, setting new benchmarks in the realm of image generation. This breakthrough promises to redefine our expectations from AI-driven creativity, particularly in the accuracy of text rendering within images. Unprecedented Accuracy in Text Rendering Historically, AI-generated images have struggled with accurately rendering text, often resulting in errors or inaccuracies that mar the final output. Ideogram 1.0 addresses this challenge head-on, boasting a nearly two-fold reduction in error [&hellip;]
- [This Week's Groundbreaking AI Use Cases](https://originshq.com/blog/this-weeks-groundbreaking-ai-use-cases/): The realm of artificial intelligence (AI) continues to evolve at a breakneck pace, offering an array of novel applications that redefine how we interact with technology. This week, we witnessed an unprecedented surge in AI integration across various platforms, from Adobe&#8217;s creative suite to innovative search engines and music generators. Let&#8217;s delve into these groundbreaking developments, showcasing how AI is reshaping the digital landscape. Adobe Integrates Generative AI into Acrobat Adobe has taken a significant step by incorporating generative AI into Acrobat, launching an AI assistant feature in beta for its standard and pro subscribers. This integration allows users to [&hellip;]
- [The Surprising Progress of AI in Video Generation: A 2024 Overview](https://originshq.com/blog/ai-in-video-generation-2024/): In early 2024, predictions about the capabilities of AI in video generation ranged from skeptical to cautiously optimistic. However, the rapid advancements in this field have surpassed expectations, leading to a reassessment of what AI can achieve in creating coherent, enjoyable video content. This blog delves into the latest developments in AI-generated video, highlighting the technologies and platforms that are redefining the landscape of digital media. AI Video Generation: Beyond Expectations Initially, many believed that AI wouldn&#8217;t be capable of producing full-length shows or films within the year. However, recent developments have proven otherwise, showcasing AI&#8217;s ability to generate videos [&hellip;]
- [Suno AI V3 Alpha: Music Generation](https://originshq.com/blog/suno-ai-v3-alpha-music-generation/): In the ever-evolving landscape of artificial intelligence and music, Suno AI has emerged as a beacon of innovation. The introduction of Suno AI V3 Alpha marks a significant milestone in text-to-song technology, offering an unparalleled experience in music creation. This latest version not only expands on the capabilities of its predecessors but also introduces a myriad of new features that set a new standard in AI-powered music generation. Inside Suno AI V3 Alpha: A Sneak Peek Suno AI V3, still in its alpha testing phase, offers an exclusive preview to pro and premiere users. This version, soon to be available [&hellip;]
- [The Evolution of Model Compression in the LLM Era](https://originshq.com/blog/llm-model-compression/): Image by author using DALL-E 3. The advent of transformers in 2017 set off a landslide of AI milestones, starting with the spectacular achievements of large language models (LLMs) in natural language processing (NLP), and quickly catalyzing advancement in other domains such as computer vision and robotics. The unification of NLP and computer vision problems into a common architecture accelerated efforts in learning joint vision-language representation spaces, which enabled the seminal achievements in vision-language modeling surrounding contrastive language-image pretraining (CLIP) in 2021, and lead to the birth of large multimodal models (LMMs). This dawning era of large models has demonstrated [&hellip;]
- [Visualisation 101: Choosing the Best Visualisation Type](https://originshq.com/blog/visualisation-101-choosing-the-best-visualisation-type/): Comprehensive guide for different visualisation use cases I believe that the primary goal of analysts is to help their product teams make the right decisions based on data. It means that the main result of analysts’ work is not just getting some numbers or dashboards but influencing reasonable data-driven decisions. So, presenting the results of our research is a critical part of analysts’ day-to-day work. Have you ever experienced not noticing some obvious anomaly until you create a graph? You are not alone. Almost nobody can extract insights from dry tables of numbers. That’s why we need visualisations to unveil [&hellip;]
- [The full training run of GPT-5 has gone live](https://originshq.com/blog/training-run-of-gpt-5/): We can expect it to be released in November, maybe on the 2nd anniversary of the legendary ChatGPT launch In similar timeframes, we will also be getting Gemini 2 Ultra, LLaMA-3, Claude-3, Mistral-2 and many other groundbreaking models (Google’s Gemini already seems to be giving tough competition to GPT-4 turbo) It is almost certain that GPT-5 will be released incrementally, these will be the intermediate checkpoints during the training of the model The actual training may take 3 months with extra 6 months for the security testing. To put GPT-5 in perspective Let us first take a look at GPT-4 [&hellip;]
- [7 Books to Read on Artificial Intelligence](https://originshq.com/blog/7-books-to-read-on-artificial-intelligence/): 2024 is thought to be the “year of AI”, where we will see even more breakthroughs than in 2023. In this post I will share some of the most interesting books about Artificial Intelligence I have been reading lately, together with my own thoughts: Life 3.0 Image by author. In this book, Tegmark talks about the 3 different tiers of life since the start of the Universe: There’s been a lot of debate about what our future, Artificial general intelligence, life 3.0 or whatever you want to call it, will look like. Few people actually believe in extreme good or [&hellip;]
- [Understanding Parameter-Efficient Finetuning of Large Language Models: From Prefix Tuning to LLaMA-Adapters](https://originshq.com/blog/understanding-peft-for-llms/): In the rapidly evolving field of artificial intelligence, utilizing large language models in an efficient and effective manner has become increasingly important. Parameter-efficient finetuning stands at the forefront of this pursuit, allowing researchers and practitioners to reuse pretrained models while minimizing their computational and resource footprints. It also allows us to train AI models on a broader range of hardware, including devices with limited computational power, such as laptops, smartphones, and IoT devices. Lastly, with the increasing focus on environmental sustainability, parameter-efficient finetuning reduces the energy consumption and carbon footprint associated with training large-scale AI models. In sum, parameter-efficient finetuning [&hellip;]
- [Some Techniques To Make Your PyTorch Models Train (Much) Faster](https://originshq.com/blog/techniques-to-make-pytorch-training-faster/): This blog post outlines techniques for improving the training performance of your PyTorch model without compromising its accuracy. To do so, we will wrap a PyTorch model in a LightningModule and use the Trainer class to enable various training optimizations. By changing only a few lines of code, we can reduce the training time on a single GPU from 22.53 minutes to 2.75 minutes while maintaining the model’s prediction accuracy. Yes, that’s a 8x performance boost! (This blog post was updated on 03/17/2023, now using PyTorch 2.0 and Lightning 2.0!) Introduction In this tutorial, we will finetune a DistilBERT model, [&hellip;]
- [Optimizing Memory Usage for Training LLMs and Vision Transformers in PyTorch](https://originshq.com/blog/optimizing-memory-usage-for-training-llms-and-vision-transformers-in-pytorch/): Peak memory consumption is a common bottleneck when training deep learning models such as vision transformers and LLMs. This article provides a series of techniques that can lower memory consumption by approximately 20x without sacrificing modeling performance and prediction accuracy. Introduction In this article, we will be exploring 9 easily-accessible techniques to reduce memory usage in PyTorch. These techniques are cumulative, meaning we can apply them on top of one another. We will begin working with a vision transformer from PyTorch’s Torchvision library to provide simple code examples that you can execute on your own machine without downloading and installing [&hellip;]
- [Improving LoRA: Implementing DoRA from Scratch](https://originshq.com/blog/improving-lora-using-dora/): Low-rank adaptation (LoRA) is a machine learning technique that modifies a pretrained model (for example, an LLM or vision transformer) to better suit a specific, often smaller, dataset by adjusting only a small, low-rank subset of the model’s parameters. This approach is important because it allows for efficient finetuning of large models on task-specific data, significantly reducing the computational cost and time required for finetuning. Last week, researchers proposed DoRA: Weight-Decomposed Low-Rank Adaptation, a new alternative to LoRA, which may outperform LoRA by a large margin. To understand how these methods work, we will implement both LoRA and DoRA in [&hellip;]
- [Finetuning Falcon LLMs More Efficiently With LoRA and Adapters](https://originshq.com/blog/finetuning-falcon-llms-more-efficiently-with-lora-and-adapters/): Finetuning allows us to adapt pretrained LLMs in a cost-efficient manner. But which method should we use? This article compares different parameter-efficient finetuning methods for the latest top-performing open-source LLM, Falcon. Using parameter-efficient finetuning methods outlined in this article, it’s possible to finetune an LLM in 1 hour on a single GPU instead of a day on 6 GPUs. Pretraining and Finetuning LLMs Before we dive into the LLM finetuning details, let’s briefly recap how we train LLMs in general. LLMs are trained in two stages. The first stage is an expensive pretraining step to train the models on a [&hellip;]
- [How to Fine-Tune LLMs with Hugging Face](https://originshq.com/blog/how-to-fine-tune-llms-with-hugging-face/): Large Language Models or LLMs have seen a lot of progress in the last year. We went from no ChatGPT competitor to a whole zoo of LLMs, including Meta AI&#8217;s Llama 2, Mistrals Mistral &amp; Mixtral models, TII Falcon, and many more. Those LLMs can be used for a variety of tasks, including chatbots, question answering, summarization without any additional training. However, if you want to customize a model for your application. You may need to fine-tune the model on your data to achieve higher quality results than prompting or saving cost by training smaller models more efficient model. This [&hellip;]
- [Understanding and Coding the Self-Attention Mechanism of Large Language Models From Scratch](https://originshq.com/blog/llm-coding-self-attention/): In this article, we are going to understand how self-attention works from scratch. This means we will code it ourselves one step at a time. Since its introduction via the original transformer paper (Attention Is All You Need), self-attention has become a cornerstone of many state-of-the-art deep learning models, particularly in the field of Natural Language Processing (NLP). Since self-attention is now everywhere, it’s important to understand how it works. Self-Attention The concept of “attention” in deep learning has its roots in the effort to improve Recurrent Neural Networks (RNNs) for handling longer sequences or sentences. For instance, consider translating [&hellip;]
- [Understanding LLM | A Transformative Reading List](https://originshq.com/blog/understanding-llm-a-reading-list/): Large language models have taken the public attention by storm – no pun intended. In just half a decade large language models – transformers – have almost completely changed the field of natural language processing. Moreover, they have also begun to revolutionize fields such as computer vision and computational biology. The following list below is meant to be read mostly chronologically, and I am entirely focusing on academic research papers. Of course, there are many additional resources out there that are useful. For example, Understanding the Main Architecture and Tasks If you are new to transformers / large language models, [&hellip;]
- [Kolmogorov Smirnov Test: When and Where To Use It](https://originshq.com/blog/kolmogorov-smirnov-test-when-and-where-to-use-it/): What is the Kolmogorov-Smirnov Test (KS test or K-S test)? The Kolmogorov Smirnov test (KS test or K-S test) is used to compare two distributions to determine if they are pulling from the same underlying distribution. In the typical ML use case, there are two distributions (A &amp; B) that you are trying to compare. One distribution A might be a feature such as, account balance, in the model training dataset. The other distribution B might be the same feature, account balance, but aggregated in a production window of time (typically daily or hourly). As a model builder, you want [&hellip;]
- [Jensen Shannon Divergence: Intuition and Practical Application](https://originshq.com/blog/what-is-jensen-shannon-divergence/): What Is JS Divergence (JS Div)? The Jensen-Shannon divergence (JS) metric – also known as information radius ( IRad ) or total divergence to the average – is a statistical measurement with a basis in information theory that is closely related to Kullback-Leibler divergence (KL Divergence) and population stability index (PSI) . The advantage of JS divergence over other metrics is mostly related to issues with empty probabilities for certain events or bins and how these cause issues with KL divergence and PSI. The JS divergence metric uses a mixture probability as a baseline when comparing two distributions as its [&hellip;]
- [KL Divergence: When To Use Kullback-Leibler divergence](https://originshq.com/blog/what-is-kl-divergence/): The basics of KL divergence and how it is used in drift monitoring What Is KL Divergence? Kullback-Leibler divergence metric is a statistical measure from information theory that quantifies the difference between one probability distribution from a reference probability distribution. KL divergence is also known as relative entropy. This post covers: KL Divergence Formula KL divergence is a non-symmetric metric that measures the relative entropy or difference in information represented by two distributions. It can be thought of as measuring the distance between two data distributions showing how different the two distributions are from each other. There is both a [&hellip;]
- [Population Stability Index (PSI): What You Need To Know](https://originshq.com/blog/what-is-psi/): Population stability index (PSI) is a statistical measure with a basis in information theory that quantifies the difference between one probability distribution from a reference probability distribution. The advantage of PSI over KL divergence is that it is a symmetric metric. PSI can be thought of as the round trip loss of entropy – the KL Divergence going from one distribution to another, plus the reverse of that. What Is Population Stability Index (PSI)? Population stability index is a symmetric metric that measures the relative entropy, or difference in information represented by two distributions. It can be thought of as [&hellip;]
- [Monitoring Text-Based Generative AI Models](https://originshq.com/blog/bleu-bert-rouge-score/): Tips for measuring text-based generative models using BLEU, ROUGE, METEOR, and Bertscore as well as prediction embeddings In recent years, text-based generative AI models have been making significant strides in natural language processing tasks such as language translation, text summarization, and dialogue generation. These models are capable of generating text that is often indistinguishable from human-generated text, making them increasingly popular in various industries, including customer service, content generation, and data analysis. While these models can be incredibly powerful and useful, they can also produce unexpected or even harmful output, making it critical to monitor them closely. For example, consider [&hellip;]
- [Normalized Discounted Cumulative Gain (NDCG)](https://originshq.com/blog/normalized-discounted-cumulative-gain-ndcg/): Overview Ranking models underpin many aspects of modern digital life, from search results to music recommendations. Anyone who has built a recommendation system understands the many challenges that come from developing and evaluating ranking models to serve their customers. While these challenges start from the data preparation and model training and continue through model development and model deployment, often what tends to give data scientists and machine learning engineers the most trouble is maintaining their ranking models in production. It is notoriously difficult to maintain models in production because of how these models are constantly changing as they adapt to [&hellip;]
- [R Squared: Understanding the Coefficient of Determination](https://originshq.com/blog/r-squared-understanding-the-coefficient-of-determination/): Introduction The R-squared metric — R², or the coefficient of determination – is an important tool in the world of machine learning. It is used to measure how well a model fits data, and how well it can predict future outcomes. Simply put, it tells you how much of the variation in your data can be explained by your model. The closer the R-squared value is to one, the better your model fits the data. In this article, we will discuss what R squared is, the math behind this metric, where it is useful and where it is not useful. [&hellip;]
- [Mean Absolute Percentage Error (MAPE): What You Need To Know](https://originshq.com/blog/what-is-mean-absolute-percentage-error/): What Is Mean Absolute Percentage Error? One of the most common metrics of model prediction accuracy, mean absolute percentage error (MAPE) is the percentage equivalent of mean absolute error (MAE). Mean absolute percentage error measures the average magnitude of error produced by a model, or how far off predictions are on average. While understanding this metric and how to calculate it is important, it’s also critical to understand its strengths and limitations when using it in production. A machine learning model may only be as good as the data used to train it, but the overall model evaluation is based [&hellip;]
- [What Is AUC?](https://originshq.com/blog/what-is-auc/): Introduction: What Is the AUC ROC Curve In Machine Learning? AUC, short for a rea u nder the ROC (receiver operating characteristic) c urve, is a relatively straightforward metric that is useful across a range of use-cases. In this blog, we present an intuitive way of understanding how AUC is calculated. How Do You Calculate AUC? To calculate AUC, we need a dataset with two columns: prediction score and actual label (i.e. Table 1). Since the actual label is binary in this case, we use +1 and -1 to denote the positive and negative classes, respectively. The important thing to [&hellip;]
- [What Is PR AUC?](https://originshq.com/blog/what-is-pr-auc/): AUC , short for area under the precision recall (PR) c urve, is a common way to summarize a model’s overall performance. In a perfect classifier, PR AUC =1 because your model always correctly predicts the positive and negative classes. Since precision-recall curves do not consider true negatives, PR AUC is commonly used for heavily imbalanced datasets where you are optimizing for the positive class. In this blog post, we will: How do you calculate PR-AUC? To calculate the PR AUC, we need a dataset with two columns: prediction score and actual label (i.e. Table 1). Since the actual label [&hellip;]
- [Calibration Curves: What You Need To Know](https://originshq.com/blog/calibration-curves-what-you-need-to-know/): In machine learning, calibration is used to better calculate confidence intervals and prediction probabilities of a given model. Calibration is particularly useful in areas like decision trees or random forests where certain classifiers only give the label of the event and don’t support native probabilities or confidence intervals. When modelers want to be confident in their predictions, they can evaluate their model via calibration to check that the predicted class distributions are similar to the current class distributions. An example is illustrative. Say a music streaming app is using a churn model to predict which users will renew or not [&hellip;]
- [Understanding and Applying F1 Score: A Deep Dive](https://originshq.com/blog/what-is-f1-score-ml/): F1 score is a measure of the harmonic mean of precision and recall. Commonly used as an evaluation metric in binary and multi-class classification, the F1 score integrates precision and recall into a single metric to gain a better understanding of model performance. F-score can be modified into F0.5, F1, and F2 based on the measure of weightage given to precision over recall depending on your use case. Why Is F1 Score Useful? There are a number of metrics useful for measuring performance of classification models. Accuracy is one of the simplest to understand. The accuracy metric measures the number [&hellip;]
- [Recall: What Is It and How Does It Differ From Precision?](https://originshq.com/blog/what-is-recall/): In machine learning, recall is a performance metric that corresponds to the fraction of values predicted to be of a positive class out of all the values that truly belong to the positive class (including false negatives). It differs from precision , which is the fraction of values that actually belong to a positive class out of all of the values which are predicted to belong to that class. Recall is also known as the true positive rate (TPR), sensitivity (SEN), probability of detection, and hit rate. Equation: Recall = predicted true positives / (true positives + false negatives) An [&hellip;]

## Pages

- [Velocity AI](https://originshq.com/products/velocity-ai/): Origins Velocity AI Suite | Enterprise AI Platform for Mid-Size Companies About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into [&hellip;]
- [AI Specialist Talent](https://originshq.com/feature/ai-specialist-talent/): AI Specialist Talent | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage [&hellip;]
- [Follow the Sun Development](https://originshq.com/feature/follow-the-sun-development/): Follow-the-Sun Development | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage Helping [&hellip;]
- [Cloud Cost Optimization](https://originshq.com/feature/cloud-cost-optimization/): Cloud Cost Optimization | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage [&hellip;]
- [Reduce Dev Costs](https://originshq.com/feature/reduce-dev-costs/): Reduce Development Costs | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage [&hellip;]
- [Faster Product Launch](https://originshq.com/feature/faster-product-launch/): Faster Product Launch | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage [&hellip;]
- [72 Hour Team Deployment](https://originshq.com/feature/72-hour-team-deployment/): 72-Hour Team Deployment | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage [&hellip;]
- [Predictive Cloud Ops](https://originshq.com/feature/predictive-cloud-ops/): Predictive Cloud Operations | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage [&hellip;]
- [AI Code Generation](https://originshq.com/feature/ai-code-generation/): AI Code Generation | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage [&hellip;]
- [AI Discovery Analysis](https://originshq.com/feature/ai-discovery-analysis/): AI Discovery Analysis | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage [&hellip;]
- [Iterative AI](https://originshq.com/products/iterative-ai/): Iterative AI Delivery | Quick Wins That Compound | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into [&hellip;]
- [Agentic Automation](https://originshq.com/products/agentic-automation/): Agentic Automation Bots | Save 1,000+ Hours Yearly | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into [&hellip;]
- [Domain LLM](https://originshq.com/products/domain-llm/): Domain-Specific LLMs | Private AI Training | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech [&hellip;]
- [AI Agents](https://originshq.com/products/ai-agents/): AI Voice Agent (24-Hour Go-Live) | Origins AI About Us Blogs Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage Helping FrontPage Scale &#8599; NuCash Building Platform at Nucash &#8599; RagaAI Building an Ai Testing Platform &#8599; Services Digital Transformation Seamless Digitization For Your Business’s Success &#8599; Data Services Building Intelligent Solutions For The Future &#8599; Artificial Intelligence Driving Innovation With Intelligent And Scalable Solutions &#8599; Cloud Redefine Scalability With Unparalleled Flexibility And Performance &#8599; DevOps Unleashing [&hellip;]
- [Products](https://originshq.com/products/)
- [Our Works](https://originshq.com/our-works/): Our Works About Us Blogs Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage Helping FrontPage Scale &#8599; NuCash Building Platform at Nucash &#8599; RagaAI Building an Ai Testing Platform &#8599; Services Digital Transformation Seamless Digitization For Your Business’s Success &#8599; Data Services Building Intelligent Solutions For The Future &#8599; Artificial Intelligence Driving Innovation With Intelligent And Scalable Solutions &#8599; Cloud Redefine Scalability With Unparalleled Flexibility And Performance &#8599; DevOps Unleashing Agility For Unprecedented Growth &#8599; Application [&hellip;]
- [We will connect with you shortly](https://originshq.com/form-submission-done/): We will get back to you shortly
- [Book a Free AI Consulting Call](https://originshq.com/book-a-call-ai/): Contact Us About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage Helping FrontPage Scale &#8599; [&hellip;]
- [Book Free Consultation Call for Mobile App Development](https://originshq.com/book-a-call-mobile/): Contact Us About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage Helping FrontPage Scale &#8599; [&hellip;]
- [Top Healthcare Consultant | Founder Friendly | Flexible Pricing](https://originshq.com/healthcare-tech-consultant/): Healthcare Software Developers About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage Helping FrontPage Scale [&hellip;]
- [Top Fintech Consultant | Founder Friendly | Flexible Pricing](https://originshq.com/fintech-consultant/): Hire Software Developemnt Company-Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Velocity AI Suite Enterprise AI platform for chat, voice, retrieval, and embedded AI &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage Helping [&hellip;]
- [dummy4testing](https://originshq.com/dummy4testing/): 72-Hour Team Deployment | Origins AI About Us Blogs Products Domain-Specific LLMs Train behind your firewall with proprietary data relevance &#8599; Agentic Automation Automate workflows and save 1,000+ staff hours yearly &#8599; Iterative AI Quick wins that compound over time with user feedback &#8599; AI Agents Intelligent agents that automate complex business processes &#8599; Our Works Samsung Samsung IoT Development and User Verification &#8599; Amazon Cost Reduction at Amazon &#8599; Sony-Playstation Building Infra at PlayStation &#8599; Yes Madam Transforming YesMadam into a Tech Company &#8599; FrontPage Helping FrontPage Scale &#8599; NuCash Building Platform at Nucash &#8599; RagaAI Building an Ai [&hellip;]
- [San Francisco Digital Transformation Company](https://originshq.com/locations/san-francisco-ca/services-digital-transformation/): Transform your business with expert digital transformation solutions in San Francisco, driving innovation and growth.
- [Charlotte Digital Transformation Company](https://originshq.com/locations/charlotte-nc/services-digital-transformation/): Transform your business with expert digital transformation solutions in Charlotte, driving innovation and growth.
- [Chicago Digital Transformation Company](https://originshq.com/locations/chicago-il/services-digital-transformation/): Transform your business with expert digital transformation solutions in Chicago, driving innovation and growth.
- [Raleigh Digital Transformation Company](https://originshq.com/locations/raleigh-nc/services-digital-transformation/): Transform your business with expert digital transformation solutions in Raleigh, driving innovation and growth.
- [Cincinnati Digital Transformation Company](https://originshq.com/locations/cincinnati-oh/services-digital-transformation/): Transform your business with expert digital transformation solutions in Cincinnati, driving innovation and growth.
- [Minneapolis Digital Transformation Company](https://originshq.com/locations/minneapolis-mn/services-digital-transformation/): Transform your business with expert digital transformation solutions in Minneapolis, driving innovation and growth.
- [Detroit Digital Transformation Company](https://originshq.com/locations/detroit-mi/services-digital-transformation/): Transform your business with expert digital transformation solutions in Detroit, driving innovation and growth.
- [Cleveland Digital Transformation Company](https://originshq.com/locations/cleveland-oh/services-digital-transformation/): Transform your business with expert digital transformation solutions in Cleveland, driving innovation and growth.
- [Houston Digital Transformation Company](https://originshq.com/locations/houston-tx/services-digital-transformation/): Transform your business with expert digital transformation solutions in Houston, driving innovation and growth.
- [Los Angeles Digital Transformation Company](https://originshq.com/locations/los-angeles-ca/services-digital-transformation/): Transform your business with expert digital transformation solutions in Los Angeles, driving innovation and growth.
- [Norfolk Digital Transformation Company](https://originshq.com/locations/norfolk-va/services-digital-transformation/): Transform your business with expert digital transformation solutions in Norfolk, driving innovation and growth.
- [Greater London Digital Transformation Company](https://originshq.com/locations/london-uk/services-digital-transformation/): Transform your business with expert digital transformation solutions in Greater London, driving innovation and growth.
- [Harrisburg Digital Transformation Company](https://originshq.com/locations/harrisburg-pa/services-digital-transformation/): Transform your business with expert digital transformation solutions in Harrisburg, driving innovation and growth.
- [York Digital Transformation Company](https://originshq.com/locations/york-pa/services-digital-transformation/): Transform your business with expert digital transformation solutions in York, driving innovation and growth.
- [Fort Myers Digital Transformation Company](https://originshq.com/locations/ft-myers-fl/services-digital-transformation/): Transform your business with expert digital transformation solutions in Fort Myers, driving innovation and growth.
- [Ann Arbor Digital Transformation Company](https://originshq.com/locations/ann-arbor-mi/services-digital-transformation/): Transform your business with expert digital transformation solutions in Ann Arbor, driving innovation and growth.
- [Lancaster Digital Transformation Company](https://originshq.com/locations/lancaster-pa/services-digital-transformation/): Transform your business with expert digital transformation solutions in Lancaster, driving innovation and growth.
- [St. Petersburg Digital Transformation Company](https://originshq.com/locations/tampa-fl/services-digital-transformation/): Transform your business with expert digital transformation solutions in St. Petersburg, driving innovation and growth.
- [Atlanta Digital Transformation Company](https://originshq.com/locations/atlanta-ga/services-digital-transformation/): Transform your business with expert digital transformation solutions in Atlanta, driving innovation and growth.
- [New Orleans  Digital Transformation Company](https://originshq.com/locations/new-orleans-la/services-digital-transformation/): Transform your business with expert digital transformation solutions in New Orleans , driving innovation and growth.
- [New York Digital Transformation Company](https://originshq.com/locations/new-york-city/services-digital-transformation/): Transform your business with expert digital transformation solutions in New York, driving innovation and growth.
- [Baltimore Digital Transformation Company](https://originshq.com/locations/baltimore-md/services-digital-transformation/): Transform your business with expert digital transformation solutions in Baltimore, driving innovation and growth.
- [Richmond Digital Transformation Company](https://originshq.com/locations/richmond-va/services-digital-transformation/): Transform your business with expert digital transformation solutions in Richmond, driving innovation and growth.
- [Washington Digital Transformation Company](https://originshq.com/locations/washington-dc/services-digital-transformation/): Transform your business with expert digital transformation solutions in Washington, driving innovation and growth.
- [Jacksonville Digital Transformation Company](https://originshq.com/locations/jacksonville-fl/services-digital-transformation/): Transform your business with expert digital transformation solutions in Jacksonville, driving innovation and growth.
- [Columbus  Digital Transformation Company](https://originshq.com/locations/columbus-oh/services-digital-transformation/): Transform your business with expert digital transformation solutions in Columbus , driving innovation and growth.
- [Toronto Digital Transformation Company](https://originshq.com/locations/toronto-ca/services-digital-transformation/): Transform your business with expert digital transformation solutions in Toronto, driving innovation and growth.
- [Boston Digital Transformation Company](https://originshq.com/locations/boston-ma/services-digital-transformation/): Transform your business with expert digital transformation solutions in Boston, driving innovation and growth.
- [Dallas Digital Transformation Company](https://originshq.com/locations/dallas-tx/services-digital-transformation/): Transform your business with expert digital transformation solutions in Dallas, driving innovation and growth.
- [Orlando Digital Transformation Company](https://originshq.com/locations/orlando-fl/services-digital-transformation/): Transform your business with expert digital transformation solutions in Orlando, driving innovation and growth.
- [Denver Digital Transformation Company](https://originshq.com/locations/denver-co/services-digital-transformation/): Transform your business with expert digital transformation solutions in Denver, driving innovation and growth.
- [Philadelphia Digital Transformation Company](https://originshq.com/locations/philadelphia-pa/services-digital-transformation/): Transform your business with expert digital transformation solutions in Philadelphia, driving innovation and growth.
- [Indianapolis Digital Transformation Company](https://originshq.com/locations/indianapolis-in/services-digital-transformation/): Transform your business with expert digital transformation solutions in Indianapolis, driving innovation and growth.
- [Pittsburgh Digital Transformation Company](https://originshq.com/locations/pittsburgh-pa/services-digital-transformation/): Transform your business with expert digital transformation solutions in Pittsburgh, driving innovation and growth.
- [San Francisco Top AI Development Company](https://originshq.com/locations/san-francisco-ca/ai-services/): Lead the future with cutting-edge AI development services in San Francisco, powering smarter solutions and automation.
- [Charlotte Top Data Engineering Company](https://originshq.com/locations/charlotte-nc/data-engineering/): Empower your business with expert data engineering services in Charlotte, building scalable and efficient data pipelines and infrastructure.
- [Charlotte Top Cloud Development Company](https://originshq.com/locations/charlotte-nc/cloud-consultant/): Unlock the full potential of cloud technologies with Charlotte’s premier cloud development services, focusing on scalability and performance.
- [Charlotte Top AI Development Company](https://originshq.com/locations/charlotte-nc/ai-services/): Lead the future with cutting-edge AI development services in Charlotte, powering smarter solutions and automation.
- [web-development](https://originshq.com/locations/san-francisco-ca/web-development/)
- [San Francisco Top Tech Consulting Company](https://originshq.com/locations/san-francisco-ca/tech-consulting/): Get strategic tech consulting services in San Francisco to accelerate your digital transformation and streamline business operations.
- [San Francisco Top Data Engineering Company](https://originshq.com/locations/san-francisco-ca/data-engineering/): Empower your business with expert data engineering services in San Francisco, building scalable and efficient data pipelines and infrastructure.
- [San Francisco Top Cloud Development Company](https://originshq.com/locations/san-francisco-ca/cloud-consultant/): Unlock the full potential of cloud technologies with San Francisco’s premier cloud development services, focusing on scalability and performance.
- [Philadelphia Top Tech Consulting Company](https://originshq.com/locations/philadelphia-pa/tech-consulting/): Get strategic tech consulting services in Philadelphia to accelerate your digital transformation and streamline business operations.
- [Philadelphia Top Data Engineering Company](https://originshq.com/locations/philadelphia-pa/data-engineering/): Empower your business with expert data engineering services in Philadelphia, building scalable and efficient data pipelines and infrastructure.
- [Philadelphia Top Cloud Development Company](https://originshq.com/locations/philadelphia-pa/cloud-consultant/): Unlock the full potential of cloud technologies with Philadelphia’s premier cloud development services, focusing on scalability and performance.
- [Philadelphia Top AI Development Company](https://originshq.com/locations/philadelphia-pa/ai-services/): Lead the future with cutting-edge AI development services in Philadelphia, powering smarter solutions and automation.
- [Indianapolis Top Tech Consulting Company](https://originshq.com/locations/indianapolis-in/tech-consulting/): Get strategic tech consulting services in Indianapolis to accelerate your digital transformation and streamline business operations.
- [Indianapolis Top Data Engineering Company](https://originshq.com/locations/indianapolis-in/data-engineering/): Empower your business with expert data engineering services in Indianapolis, building scalable and efficient data pipelines and infrastructure.
- [Indianapolis Top Cloud Development Company](https://originshq.com/locations/indianapolis-in/cloud-consultant/): Unlock the full potential of cloud technologies with Indianapolis’s premier cloud development services, focusing on scalability and performance.
- [Indianapolis Top AI Development Company](https://originshq.com/locations/indianapolis-in/ai-services/): Lead the future with cutting-edge AI development services in Indianapolis, powering smarter solutions and automation.
- [Pittsburgh Top Tech Consulting Company](https://originshq.com/locations/pittsburgh-pa/tech-consulting/): Get strategic tech consulting services in Pittsburgh to accelerate your digital transformation and streamline business operations.
- [Pittsburgh Top Data Engineering Company](https://originshq.com/locations/pittsburgh-pa/data-engineering/): Empower your business with expert data engineering services in Pittsburgh, building scalable and efficient data pipelines and infrastructure.
- [Pittsburgh Top Cloud Development Company](https://originshq.com/locations/pittsburgh-pa/cloud-consultant/): Unlock the full potential of cloud technologies with Pittsburgh’s premier cloud development services, focusing on scalability and performance.
- [Pittsburgh Top AI Development Company](https://originshq.com/locations/pittsburgh-pa/ai-services/): Lead the future with cutting-edge AI development services in Pittsburgh, powering smarter solutions and automation.
- [Charlotte Top Tech Consulting Company](https://originshq.com/locations/charlotte-nc/tech-consulting/): Get strategic tech consulting services in Charlotte to accelerate your digital transformation and streamline business operations.
- [Washington Top Cloud Development Company](https://originshq.com/locations/washington-dc/cloud-consultant/): Unlock the full potential of cloud technologies with Washington’s premier cloud development services, focusing on scalability and performance.
- [Washington Top AI Development Company](https://originshq.com/locations/washington-dc/ai-services/): Lead the future with cutting-edge AI development services in Washington, powering smarter solutions and automation.
- [Jacksonville Top Tech Consulting Company](https://originshq.com/locations/jacksonville-fl/tech-consulting/): Get strategic tech consulting services in Jacksonville to accelerate your digital transformation and streamline business operations.
- [Jacksonville Top Data Engineering Company](https://originshq.com/locations/jacksonville-fl/data-engineering/): Empower your business with expert data engineering services in Jacksonville, building scalable and efficient data pipelines and infrastructure.
- [Jacksonville Top Cloud Development Company](https://originshq.com/locations/jacksonville-fl/cloud-consultant/): Unlock the full potential of cloud technologies with Jacksonville’s premier cloud development services, focusing on scalability and performance.
- [Jacksonville Top AI Development Company](https://originshq.com/locations/jacksonville-fl/ai-services/): Lead the future with cutting-edge AI development services in Jacksonville, powering smarter solutions and automation.
- [Columbus  Top Tech Consulting Company](https://originshq.com/locations/columbus-oh/tech-consulting/): Get strategic tech consulting services in Columbus  to accelerate your digital transformation and streamline business operations.
- [Columbus  Top Data Engineering Company](https://originshq.com/locations/columbus-oh/data-engineering/): Empower your business with expert data engineering services in Columbus , building scalable and efficient data pipelines and infrastructure.
- [Columbus  Top Cloud Development Company](https://originshq.com/locations/columbus-oh/cloud-consultant/): Unlock the full potential of cloud technologies with Columbus ’s premier cloud development services, focusing on scalability and performance.
- [Columbus  Top AI Development Company](https://originshq.com/locations/columbus-oh/ai-services/): Lead the future with cutting-edge AI development services in Columbus , powering smarter solutions and automation.
- [Toronto Top Tech Consulting Company](https://originshq.com/locations/toronto-ca/tech-consulting/): Get strategic tech consulting services in Toronto to accelerate your digital transformation and streamline business operations.
- [Toronto Top Data Engineering Company](https://originshq.com/locations/toronto-ca/data-engineering/): Empower your business with expert data engineering services in Toronto, building scalable and efficient data pipelines and infrastructure.
- [Toronto Top Cloud Development Company](https://originshq.com/locations/toronto-ca/cloud-consultant/): Unlock the full potential of cloud technologies with Toronto’s premier cloud development services, focusing on scalability and performance.
- [Toronto Top AI Development Company](https://originshq.com/locations/toronto-ca/ai-services/): Lead the future with cutting-edge AI development services in Toronto, powering smarter solutions and automation.
- [Boston Top Tech Consulting Company](https://originshq.com/locations/boston-ma/tech-consulting/): Get strategic tech consulting services in Boston to accelerate your digital transformation and streamline business operations.
- [Boston Top Data Engineering Company](https://originshq.com/locations/boston-ma/data-engineering/): Empower your business with expert data engineering services in Boston, building scalable and efficient data pipelines and infrastructure.
- [Boston Top Cloud Development Company](https://originshq.com/locations/boston-ma/cloud-consultant/): Unlock the full potential of cloud technologies with Boston’s premier cloud development services, focusing on scalability and performance.
- [Boston Top AI Development Company](https://originshq.com/locations/boston-ma/ai-services/): Lead the future with cutting-edge AI development services in Boston, powering smarter solutions and automation.
- [Dallas Top Tech Consulting Company](https://originshq.com/locations/dallas-tx/tech-consulting/): Get strategic tech consulting services in Dallas to accelerate your digital transformation and streamline business operations.
- [Dallas Top Data Engineering Company](https://originshq.com/locations/dallas-tx/data-engineering/): Empower your business with expert data engineering services in Dallas, building scalable and efficient data pipelines and infrastructure.
- [Dallas Top Cloud Development Company](https://originshq.com/locations/dallas-tx/cloud-consultant/): Unlock the full potential of cloud technologies with Dallas’s premier cloud development services, focusing on scalability and performance.
- [Dallas Top AI Development Company](https://originshq.com/locations/dallas-tx/ai-services/): Lead the future with cutting-edge AI development services in Dallas, powering smarter solutions and automation.
- [Orlando Top Tech Consulting Company](https://originshq.com/locations/orlando-fl/tech-consulting/): Get strategic tech consulting services in Orlando to accelerate your digital transformation and streamline business operations.


[comment]: # (Generated by Hostinger Tools Plugin)