XIO

xWire

Posts from every feed QUEST has found across the web — blogs, news, podcasts, channels and more.

DeepLearning.AI - Latest topics

Latest topics

C2W1 Assignment 1 – Empty PROJECT Folder and Unresponsive Terminal

Hi @magdalena.bouza, I’m experiencing a technical issue with the Coursera lab for the DeepLearning.AI Data Engineering Professional Certificate. Course: Source Systems, Data Ingestion, and Pipelines Assignment: Course 2, Week 1 – Assignment 1: Troubleshooting Database Connectivity on AWS Issue: The VS Code lab interface loads, but the environment is not functioning properly. The PROJECT folder is empty, and no assignment files appear. The integrated terminal opens but remains blank and does not…

Introducing TPDZI ATLAS: Evidence-Governed AI for Research and Engineering

Hello everyone! I’m an independent researcher and the creator of TPDZI ATLAS, an AI-assisted multidisciplinary research and engineering platform currently under development. My main interest is not simply making AI generate better answers, but building systems where AI can coordinate specialized tools, work with scientific evidence, preserve uncertainty, and support complex engineering investigations without losing human oversight. ATLAS brings together multiple research directions, including:…

500 error in L5

I am using the cloud hosted setup. I am able to bring up the MCP client fine, and it lists the right tools. But putting any query in the chatbot gives error 500. Here are the messages from the terminal: (mcp_project) jovyan@ip-172-29-95-38:~/work/L5/mcp_project$ uv run mcp_chatbot.py Anthropic: ANTHROPIC_API_KEY and ANTHROPIC_AUTH_TOKEN are both set. Using ANTHROPIC_API_KEY. Processing request of type ListToolsRequest Connected to server with tools: [‘search_papers’, ‘extract_info’] MCP Chatbot…

Automatically screening AI-generated video clips before human review. What’s working for you?

Hi everyone, I produce a lot of short AI-generated video clips, and manual review has become the bottleneck. Most rejects fail for the same few reasons: warped hands, flickering backgrounds, a character’s face drifting mid-clip, or motion that breaks physics. I’d like an automatic first pass that throws out the obvious failures, so a human only watches the clips that have a chance. A few things I’m unsure about: Has anyone had good results using a vision-language model as a judge on sampled…

Re-submit fails with “An unexpected error occurred” even though all public tests pass

Hi DeepLearning.AI team, I’m having a submission/platform issue with the code assignment. Program/Course: Generative AI for Software Development — Course 2 Module: Module 3: Dependency Management Graded item: “Solving Versioning and Dependency conflicts with an LLM” (Code Assignment) In the Coursera lab, my solution appears correct locally: python -B unittests.py → All tests passed! python -c "from magic_summation_solution import magic_summation; print(magic_summation(30, seed=10))" → 46 python…

How Should RAG Systems Handle Outdated or Conflicting Information?

When building a Retrieval-Augmented Generation (RAG) system, retrieving relevant documents is only part of the challenge. Sometimes the knowledge base contains outdated information or multiple documents that contradict each other. I’m curious how developers handle these situations in practice. For example, should a retrieval system prioritize documents based on publication date, source reliability, or some combination of factors? How do you prevent an LLM from confidently generating an answer…

Managing context like Git

Hey guys, today I wanted to share my thoughts on context engineering. The core idea is treating context (ctx) as the basic unit—like a node—and managing it just like Git. You can branch it, and you can merge it. Our minds don’t think linearly, so why should our chat boxes be strictly linear? That’s why I built the GGUF_Libra project. Memory Mechanism The problem with RAG is that it only fishes up raw strings, not structured cause-and-effect chains. Because of this, the model gets confused—it…

AI Grading problem

Add a clear title Grading Probelm Add all required tags in the tag option (top-right of the post): ( dl-ai-learning-platform or Coursera-platform.) The course week/module Course - Sequence Models Module - week 1 - Programming assignment - Building your Recurrent Neural Network - Step by Step , Dinosaur Island-Character-Level Language Modeling , Jazz Improvisation with LSTM . Module week 2 - Programming assignment - Operations on Word Vectors - Debiasing , Emojify Module week 3 -Programming…

Tokatlas: looking for real workloads to test multi-model API cost and compatibility

Hi everyone — I’m on the Tokatlas team, a commercial multi-model API platform. We’re inviting developers to test it on a small, real workload and share the results, including failed integrations. A lower token rate isn’t enough if a model needs more retries or its output fails validation. For this trial, we’d like to compare cost per accepted result, alongside response time and output quality. A proposed test would keep the input set, prompt, output checks and retry limit fixed, then record:…

Module 2, ungraded lab

3.2. Step 2 — Execute Code and Create Chart --------------------------------------------------------------------------- KeyError Traceback (most recent call last) File /usr/local/lib/python3.11/site-packages/pandas/core/indexes/base.py:3805, in Index.get_loc(self, key) 3804 try: -> 3805 return self._engine.get_loc(casted_key) 3806 except KeyError as err: File index.pyx:167, in pandas._libs.index.IndexEngine.get_loc() File index.pyx:196, in pandas._libs.index.IndexEngine.get_loc() File…

Counting "agentic" vs "generative AI" on 5,305 earnings call transcripts, and the ASR traps

I work at Equibles, where we just published a count of how S&P 500 companies talk about AI on their earnings calls. Sharing the method and the traps, because they apply to any keyword study on speech-to-text data. Setup. 5,305 earnings calls held by 494 current S&P 500 companies from 5 January 2024 to 23 October 2026, speaker-by-speaker transcripts, grouped by the calendar quarter in which each call was held. Results “Agentic” went from 0.2% of calls in January to March 2024 to 20.3% in July to…

AI terminology

Hi there! I had a question regarding AI terminology. I dont quite understand the difference between machine learning, data science, and deep learning. Like, do all of these terminologies fall under Generative AI. This is the conclusion I have come up with, and I want to know if it is true or not. Data science uses both machine learning and deep learning, and is best with structured and unstructured data. Machine learning is best with structured data. The difference between data science and…

Course is loading the wrong lab

Today, I tried to complete my pending lab work. Unfortunately, I am unable to complete Assignment 3, Week 4 of the Data Engineering course on Coursera because the lab is loading the wrong materials. Specifically, it is showing the material for Assignment 1, Week 1 instead. For your reference, I have attached a screenshot of the issue. Thank you for your understanding and assistance. 4 posts - 4 participants Read full topic

CAEL 0.3: exploring a non-neural assistant with symbolic helpers

I’m building CAEL and would like feedback on how to evaluate a hybrid assistant whose inference uses no neural networks, transformers or external language models. CAEL combines counted language models, cellular automata, symbolic reasoning, exact arithmetic and local retrieval. The question behind the project is how useful an assistant can become by combining a small statistical language component with specialized, inspectable helpers. The public CAEL 0.3 runtime includes chat, supported…

Assignment 5: Capstone Project Part 2 Stuck on loading screen

Assignment 5: Capstone Project Part 2 - Data Quality and Orchestration and Visualization is stuck on loading screen, I tried opening multiple times using different browsers, different machines multiple times and it is still the same. I already made a lot of progress on the lab but it is not opening when I’m trying now. Below is what I see: 9 posts - 4 participants Read full topic

Generative AI with Large Language Models | Lab not accessible

Lab 1 - Generative AI Use Case: Summarize Dialogue I am unable to acces the lab and seeing following message in the screen "Please switch to the latest session of this course. Visit the DeepLearning.AI Forum for more info. " . Please let me know how can i switch latest session of this course inorder to access the Lab " 2 posts - 2 participants Read full topic

Not able to submit Module 2 assigntment even my implementation passed all tests

Before you post, please make sure to: Add a clear title that describes your question. Add all required tags in the tag option (top-right of the post): The platform (e.g., dl-ai-learning-platform or Coursera-platform.) The course week/module Write a detailed description of your issue. Note: Screenshots are welcome, but please do not post full assignment code. For more information, please check the FAQ category. *Delete this message before posting 1 post - 1 participant Read full topic

Module 2 Graded (and Ungraded) assignments failing due to deprecated model o4-mini (deprecated today 10/23/26)

I started working through the agentic AI course and discovered that the Module 2 graded and ungraded lab uses o4-mini which was deprecated today (Oct 23) Right now you can not submit your lab without it failing OpenAI Developers Deprecations | OpenAI API Find information about OpenAI API deprecations and recommended replacements. 1 post - 1 participant Read full topic

How to Download Notebook Content w/ Chrome

This course is incredible, but I wanted to see the code in my own IDE. I tried following advice from this post, but it did not work due to a limitation apparently in Google Chrome: Downloading your Notebook, Downloading your Workspace and Refreshing your Workspace This script works for me. Obviously run this at your own risk: # Build a backup tarball of this folder and show a download link that works # inside the embedded course notebook (opens the file in a new top-level tab, # because Chrome…

Unable to access week 3 assignment

Any time I try to access the Week 3 assignment for the Introduction to Data Engineering course, when I open the lab, it always takes me to the Week 2 assignment, as you can see here. If you look at the top, Assignment 3 is there, but the course content all shows Week 2 work. Get the URL. It says week 3, but the assignment is for week 2. 3 posts - 3 participants Read full topic

Reflective Research Agent demo stuck on “Generating steps for the search” for 10+ minutes

Platform: dl-ai-learning-platform or Coursera-platform, whichever you’re using Course week/module: the module that contains this activity (the course covers the research agent in the first module, so check your syllabus for the exact name) Hi everyone, I’m having trouble with the “Reflective Research Agent: Planning and Tool Use” ungraded activity. After I submit my prompt, the page shows “ Generating steps for the search:” followed by my prompt, but nothing else appears. No plan, search steps,…

Staff Announcement: All Vocareum Labs are available again

Hi everyone! We are happy to announce that all Data Engineering labs are once again available, except for Course 3 Week 2 Assignment, which we’re working on. Please, if you run into any issues create a new thread and tag me (@magdalena.bouza) so we can solve them ASAP. We apologize for all the inconveniences, and thank you all for your patience during the process. 16 posts - 6 participants Read full topic