Welcome to APC Mastery Path - Your Road to Success in RICS APC and Beyond!
Are you an RICS APC candidate looking to ace your assessment and achieve your professional goals? Interested in harnessing the power of AI in the construction industry? Look no further!
APC Mastery Path provides top-notch education, expert guidance, and invaluable insights in: - RICS APC Excellence - AI in Construction - Industry Insights - Expert Mentoring - Networking
Join us on APC Mastery Path and embark on your journey towards success. Subscribe today and start exploring the path to a brighter future! Subscribe now!
@@Maisonier You can still use Open WebUI for large files. The time it is going to take to process PDFs is going to be dependent on your machine GPU, CPU & RAN. You can also check how to use marker-pdf package to extract text data from pdfs and then you can use the extracted text.md file to feed in any RAG system. Hope this helps a bit.
Another issue is that when trying to load the pre-trained model is not loading. It is showing me something like: "Error loading pretrained model: Error parsing message"? Need help please.
Merci beaucoup, chérie! Ça me fait tellement plaisir de savoir que tu trouves mes efforts utiles. J'adore partager et je suis toujours là pour offrir plus! 💖
Many thanks Soha for your comment and support. Hope you found it useful and deployable at your end. I am currently working on a number of other ideas as well, so stay tuned. 😉
🎯 Key Takeaways for quick navigation: 00:00 🚀 *Introduction to Llava LLM and its visual capabilities* - Overview of Llava Large Language Model (LLM) and its ability to understand and describe images. - Discussion on deploying Llava LLM for image understanding and text deciphering. - Exploration of the model's architecture and the process of building a chatbot utilizing its capabilities. 01:13 📦 *Importing necessary items for the system architecture* - Importing images, Python libraries, and Llava large language model. - Utilizing Python libraries for image conversion and creating a prompting function. - Description of the direct application of the system, converting images into tensors, and providing prompts for image description. 02:11 💻 *Development environment and required packages* - Overview of the integrated development environment (IDE) and Python version used. - Essential Python packages such as Glob, Requests, Transformers, Torch, Accelerate, and Llava Torch. - Recommended specifications for hardware requirements, including CPU and GPU specifications. 04:19 📚 *Importing necessary packages and modules* - Importing required packages and modules for image processing and model loading. - Steps to disable torch where conflicts arise and importing the model, images, and tokenizer. - Verification of model and image imports and tokenizer initialization. 05:05 🖼️ *Processing images and initializing the tokenizer* - Processing images into tensors using the Llava Torch library. - Downloading and importing the model from the Hugging Face repository. - Initializing the tokenizer and enabling quantization for model usage on consumer-grade GPUs. 07:35 🔍 *Creating a function to process images* - Defining a function to convert input images into tensors understandable by Llava LLM. - Specifying image processing parameters and device for GPU usage. - Demonstrating the image processing function with an example image and output tensor. 08:43 💬 *Converting prompts for model inference* - Defining conversational mode and creating a function to convert prompts into a format understood by the Llava LLM. - Importing conversational mode from pre-made templates on the Hugging Face website. - Applying the prompt conversion function and demonstrating with example prompts and outputs. 10:50 ⚙️ *Creating an overarching function for model interaction* - Defining a function to interact with the model, providing image inputs, prompts, and receiving answers. - Processing images and prompts, defining input IDs, and setting inference parameters. - Demonstrating the overarching function with image and prompt inputs and receiving text-based outputs. 13:29 ⏱️ *Analyzing model inference time and output accuracy* - Evaluating inference time for image processing and prompt conversion. - Assessing the accuracy of model outputs in describing images based on given prompts. - Calculating and comparing inference times for different model tasks on a consumer-grade PC. 15:22 📃 *Extending functionality to optical character recognition (OCR)* - Expanding Llava LLM capabilities to include text recognition from images. - Importing images containing text and querying for titles and summaries. - Assessing model accuracy in summarizing text from images and showcasing real-world applications. 17:12 📚 *Conclusion and additional resources* - Summary of video content and invitation to explore further resources. - Promoting an online guide on querying images using Llava LLM available on the website. - Mentioning additional services and tools offered on the APC Mastery Path website, including mentoring, study progress tracking, and AI deployment guidelines in the construction industry.
Yeah I know about Ollama, I always love to get my hands dirty by exploring the model from scratch and build things from the ground up. Many thanks for your fruitful contribution.