LLM
Hello! In this article, we take a detailed look at how image token counts are calculated when using Claude 4.5 Sonnet/Haiku and Claude 4.1 Opus via the API. How Image Token Counts Are Calculated Images sent to the Claude 4.5 API are counted as tokens, just
Python
Hello! When writing custom classes in Python, have you ever run into the baffling situation where "the object clearly exists, but an if statement judges it False" ? In this article, we explain the unexpected truthiness behavior that arises when you implement the __len__ method! A Bug We Actually
News
Qualiteg Inc. will exhibit at CEATEC 2025, held at Makuhari Messe from Tuesday, October 14 to Friday, October 17, 2025. At our booth, we will showcase our photorealistic AI avatar technology "MotionVox🄬" along with our latest AI technologies and business innovation solutions. Exhibition Overview * Dates: Tuesday, October 14
Daily Dev Tips
Hello! In this post, we look at how to resolve the "libcudnn.so.9: cannot open shared object file" error that can occur during GPU inference with ONNX Runtime. When running GPU inference with ONNX Runtime, you may run into a CUDA provider initialization error. This article explains
IT & AI Technology
Hello! In this third installment of the series, we cover joining clients and servers to the domain. Introduction Hello, and welcome to Part 3 of the series, "Joining Clients and Servers to the Domain." In the previous installment (Part 2), we walked through building an Active Directory domain
Generative AI Frontlines
Hello — this is the consulting division of Qualiteg Inc. The rapid evolution of generative AI is fundamentally reshaping how software is built. In 2024, large language models such as Claude, GPT-4, and Gemini dramatically improved their code generation capabilities, and development tools like GitHub Copilot, Cursor, and Windsurf came into
PyTorch
Hello! In this post, we explain how to deal with the following problem in a PyTorch 2.6.0 environment. NVIDIA GeForce RTX 5090 with CUDA capability sm_120 is not compatible with the current PyTorch installation. The current PyTorch install supports CUDA capabilities sm_50 sm_60 sm_70
Daily Dev Tips
Hello! OpenCV is widely used in image processing and video analysis. When working with video processing in production, however, you may occasionally run into an error like the following. cv2.error: OpenCV(4.11.0) /io/opencv/modules/imgcodecs/src/loadsave.cpp:929: error: (-215:Assertion failed) !_img.empty() in
AI Mathematics
Hello! Our lipsync technology series is finally entering its home stretch. Last time (Part 4), we took a detailed look at the LSTM training process and its limits. While we saw the LSTM's strength of learning effectively even from limited data, it also became clear that it struggles
IT & AI Technology
Hello! In this second installment of the series, we walk through building a domain environment — the steps for setting up a test lab. Series Overview Chapter 1: Understanding the Basics - Fundamentals of Active Directory and Kerberos/NTLM authentication [★This installment★] Chapter 2: Building the Domain Environment - Steps for
Daily Dev Tips
Hello! Today's topic is the "closure problem." Have you ever defined a function inside a loop in Python? If so, you may have run into some rather puzzling behavior. In this article, we explain the "closure problem" — something every Python programmer runs into at
News
We are pleased to share that our "Qualiteg Blog" has been featured in "トレンドをキャッチアップ！AIに関する情報が得られるメディア・ブログまとめ" (Catch Up on the Trends! A Roundup of Media and Blogs for AI Information), an article published by Freelance Hub, a project-search service for freelancers in Japan. About the feature
LLM
Hello! OpenAI's vision-capable models (that is, LLMs that accept image input) use two different methods for converting images into tokens. The latest GPT-5 and GPT-4.1 families introduced a patch-based method that differs from the traditional tile-based approach. This change significantly improves the efficiency of image processing and
Daily Dev Tips
Hello! This is the Qualiteg Product Development team! This is the story of how simply converting a synchronous method into an asynchronous one (async) hit us with an unexpected memory leak*. It happened while we were developing a video processing system that uses deep learning models. We had a requirement
LLM Security
Hello! Today we take a close look at "Zero Trust," a new way of thinking about security, and at AI security centered on LLMs. Introduction Three Paradigm Shifts Are Happening at Once Right now, three major transformations are underway in corporate IT environments. The first is a shift
AI Mathematics
1. Position Loss (L_position) - Accuracy of the Mouth Shape Time Mouth opening Truth Predicted L_position = Σᵢ wᵢ × ||y_pred - y_true||² Computes the difference between predicted and ground-truth values at each time step. Important parameters (jaw opening, mouth opening) receive larger weights. jaw_open: ×2.0