NCA-GENL practice questions
NVIDIA · NCA-GENL · 300 questions
Validates foundational skills for developing, integrating, and maintaining LLM-based AI applications using NVIDIA solutions, covering core ML and AI knowledge, software development with Python and LLM frameworks, experimentation with transformer-based models, GPU-accelerated data analysis, and trustworthy AI principles.
This course contains the use of artificial intelligence.
About the NCA-GENL exam
- Exam fee
- $125 USD
- Time allowed
- 1 hour
- Questions
- 50-60
- Format
- Online, remotely proctored, multiple-choice
Exam details published by the vendor, checked 28 August 2026. Vendors change fees and formats without notice — confirm on the vendor's own page before you book.
Practice Quizzes
Test your knowledge with standard 20-question practice sets.
Quiz 1
Quiz 2
Quiz 3
Quiz 4
Quiz 5
Quiz 6
Quiz 7
Quiz 8
Quiz 9
Quiz 10
Quiz 11
Quiz 12
Quiz 13
Quiz 14
Quiz 15
Browse by Domain
Study specific topics at your own pace.
Core Machine Learning and AI Knowledge · 90 questions
- A recycling depot already scores bin weight with a small fitted predictor. A vendor then offers a many-layer network on camera frames of the sorting line, and a third pitch offers a text model that drafts weekly recycling memos. How are classical machine learning, deep learning, and a generative LLM related?
- A dock-permit booth has six numeric columns and a few hundred labeled sailings. A consultant insists they must train a large neural network “because AI.” What should they do first?
- An oyster-farm intern draws a single unit that multiplies incoming measurements by learned numbers, adds a bias, and then squashes the result before passing it on. What are the learned numbers and the squash step for?
- A cider-press workshop trains a grade predictor and watches a single number fall as predicted juice acidity gets closer to lab truth. What is that number?
- During one training step a foghorn-log classifier first produces a guess from the current weights, then uses the mismatch to nudge those weights. Which half is which?
- A fire-lookout hears that training “walks downhill on the error surface” by taking many small steps instead of solving for weights in one closed equation. Which idea is that?
- An herbarium model memorizes last decade’s catalog notes and then fails on this year’s new sheets, while a second model is so simple it misses obvious moisture patterns even on the training years. What are those two failures?
- A drawbridge office splits historic opening logs three ways: one pile to fit weights, one pile to choose among a few training recipes, and a locked-away pile used once at the end. What is each pile for?
- A bird-banding station spends a rainy week fitting a species tagger on archived photos, then the tagger later labels each new photo in seconds on a dock tablet. Which span is which?
- A trolley-photo archive notices that stacking layers of matrix multiplies crawls on a laptop CPU but finishes far sooner when the same work is sent to a GPU. Why does that hardware help deep training?
- A weather-balloon shed is told “the training job uses CUDA.” An intern asks whether they must write a custom GPU kernel. What should they understand?
- A lighthouse postcard archive has only a few dozen labeled captions but can start from a network already trained on a huge public text collection and adapt the last stages. What is that approach?
- A maple-sugar cooperative’s training log shows how many times the full note set was seen, how many notes were used per update, and how large each weight nudge was. Which three knobs are those?
- A swing-bridge operator can predict delay minutes from five numeric sensors, or they can draft a passenger announcement from free-text incident notes. Which model family fits which job?
- A municipal vineyard desk wants three jobs: flag blight or no blight, predict harvest tons as a number, and group unlabeled blocks of vines that behave alike. Which learning types are those?
- A flour-mill lab has mixed numeric and categorical columns and needs a strong off-the-shelf predictor. Which family fits that tabular science problem?
- A ranger first-aid post has 400 rows of lab columns and no images. One vendor sells a large language model; another offers an XGBoost model on the same sheet. Which should they choose?
- A creek-gauge office must tell the council which sensors most influenced a flood-risk score. Which method can surface that contribution?
- A marina salvage sheet has empty cells and a mix of counts and category codes. Why are boosted trees often used on that kind of table?
- A towpath museum gets four pitches: a linear model for a roughly straight fee-versus-headcount line, clustering for unlabeled visitor paths, boosted trees for a messy mix of ticket fields, and a language model for rewriting exhibit cards. How should they match those pitches?
- A county assessor hears that XGBoost can run faster on a GPU. What does that mean at associate depth?
- A lighthouse supply desk compares one shallow tree that splits on a single weather flag with a sequence of trees that each correct the previous residual. What is the second pattern?
- A harbor-pilot office wants to study which docks connect through shared tug assignments. What is the natural shape of that data?
- A volunteer trolley museum has a few hundred stops and transfer links and already works in Python on a laptop. Which toolkit fits that modest graph work?
- The same connectivity study now spans tens of millions of transfer links and must finish in a short window. Where should the graph analytics run?
- A river-freight desk asks three questions: the shortest tow route between locks, which lock is a bottleneck hub, and which clusters of landings trade mostly among themselves. Which graph-task families are those?
- A lock museum wants an assistant that can follow “this exhibit was restored by the same shop that built that engine.” Why prefer a graph of entities and relations over a single FAQ spreadsheet?
- Three GPU jobs land on one desk: wrangle a giant CSV, train a classical GPU model, and run community detection on a huge network. Which RAPIDS libraries match those jobs?
- A canal-archive team used to feed tokens one after another through a recurrent net and lost track of early tokens. Why did transformers become the building block of modern LLMs?
- A lock log says “the lower gate jammed after the upper gate opened.” The model must let “jammed” look at “lower gate,” not just nearby filler words. Which mechanism scores those relationships?
- An intern describes attention as asking a question about the current token, matching that question against other tokens, and then gathering useful content from the matches. Which three jobs are those?
- One head of a ferry-schedule model tracks who-did-what, another tracks times, and another tracks places. Why use several attention heads instead of one averaged view?
- The notices “crew relieved captain” and “captain relieved crew” would look the same if the model only saw an unordered bag of token vectors. What restores order?
- A tide-table caption is split into pieces that may be whole words or fragments before anything is embedded. What are those pieces?
- After tokenization, each piece becomes a list of numbers so that “skiff” and “dinghy” land nearer than “skiff” and “invoice.” What are those vectors?
- A lighthouse keeper pastes three seasons of logs into a chat box and the tail of the prompt is silently dropped. What limit was hit?
- Three tools sit on a docent’s bench: one reads a whole plaque and labels it, one writes the next sentence of a story, and one reads a source language then writes a target language. Which architecture families are those?
- While writing the next word of a lock announcement, a decoder must not already see the words it has not written yet. Which rule enforces that?
- A plaque-tagging model can let “closed” look at both “temporarily” on its left and “for painting” on its right. What kind of attention is that?
- After tokens have mixed information through attention, each position still goes through the same small multilayer network of its own. What is that sublayer?
- A deep stack of transformer blocks would be brittle if each block replaced its input outright. Which associate-level tricks keep that stack trainable?
- A ferry chatbot emits one piece, then feeds that piece back in to choose the following piece, and so on. What is that loop?
- During pretraining a transformer can score many token positions of a packed sequence in one pass, but at serving time a decoder still emits tokens one after another. How do those time-shapes differ?
- A bilingual exhibit card is read by one stack, then a second stack writes the other language while looking back at the first stack’s states. What is that looking-back?
- A canal-history project cannot afford librarians to tag every word in a century of minutes, yet it can hide some words and ask the model to recover them from the rest. What is that approach?
- Training blanks out a few tokens in a lock log (“The gate failed”) and scores whether the model fills the blank using both sides. Which objective is that?
- Another recipe never shows the future: the model always predicts the next word of a captain’s memoir given only the words so far. Which objective is that?
- A plaque office needs bidirectional understanding for tagging and similarity, not open-ended story writing. Which official LLM variant fits?
- A research desk hears that some modern LLMs are scaled transformer variants associated with NVIDIA’s Megatron line. How should they treat that name on this exam?
- First a model absorbs generic lock-and-harbor language from a huge unlabeled pile; later a small labeled set of “delay / no delay” notes adapts it to one desk’s ticket codes. What are those two stages?
- Early transformers still needed costly labeled corpora for each task. Why did self-supervision make BERT-, Megatron-, and similar variants so effective?
- A town council asks why a “large” language model needs huge text, many weights, and GPU time. What associate picture should they get?
- A raw next-token model completes “The lower gate” with more memoir text; after an instruction stage it instead follows “Summarize this lock log in two bullets.” What did that stage teach?
- Two variants sit on a shelf: a BERT-like encoder for tagging exhibit text and a large decoder-style Megatron-family model for drafting new copy. How should they pick?
- A hut-warden chatbot starts giving rambling answers. The team rewrites the request and adds an example instead of opening a training job. What did they change?
- First they just ask “label this incident.” Quality is uneven, so they paste three labeled incidents above the new one. What are those two patterns?
- A ferry-desk prompt is too vague, then too strict, then finally states format and audience. How should prompt work be treated?
- A museum bot should always stay in the docent voice and never invent exhibit years, while each visitor question changes. How should those pieces be split?
- A lock-fee word problem fails when the model jumps to a total, then succeeds when the prompt requires stepwise arithmetic in plain language. What prompt pattern is that?
- The same log dump must become a three-bullet brief, a list of equipment names, or a visitor Q&A. What should the three prompts do?
- A well-written instruction is buried behind an entire policy binder pasted into the prompt, and the model never sees the last chapter. What went wrong?
- A draft exhibit label is dull at a low randomness setting and wildly off-brand at a high one. What kind of controls are those?
- One failure is “tell me about the gate” with no which-gate or what-format; another is asking a small model to recite a regulation that was never in its training or prompt. What is the difference?
- A ticket office wants a bot that answers only from the pasted timetable and says when the timetable is silent. What should the prompt include?
- “Lower gate will not seat” and “the downstream leaf failed to close” should score as similar even though they share few words. Which job is that?
- A museum encodes every catalog card into a vector so a new visitor question can find the nearest cards. What is that encoding step?
- Given a lock-operation paragraph and the question “What time did the gate reopen?”, the system highlights a span already in the paragraph. What kind of QA is that?
- A new “ice-warning” ticket type appears and there is no trained head for it yet. An encoder is asked to pick among label names it was never fine-tuned on, using the meaning of those names. What is that?
- A harbor log must mark each ship name, berth, and clock time inside the sentence. Which encoder setup fits?
- Each visitor comment should become one of {praise, complaint, question}. Which encoder job is that?
- A dock office only needs a three-way tag on each radio transcript and must stay fast and on-device. Why prefer a small encoder classifier over a large generative LLM?
- After encoding, “capstan” sits near “winch” and far from “cafeteria menu.” Why does semantic search work?
- The same decoder writes a stern safety bulletin when primed with a safety template, and a cheerful kids’ tour script when primed with a tour template. What is doing the steering?
- A lock desk needs three named fields back — incident, equipment, delay minutes — not a paragraph. How should they condition the decoder?
- Staff paste a raw incident paragraph and ask for a two-sentence public notice that does not add facts. What is the source text doing?
- A visitor asks “and the Sunday one?” which only makes sense after the previous turn about winter ferries. What must condition this decoder turn?
- A decoder is asked to turn a short weather code into a spoken-style script, or a table of sailings into prose. What associate “modality” claim is fair?
- The only required output is open or closed for a gate-status feed. Should they call a long-form decoder?
- A helpful lock-bot answers every question, including how to force a gate against procedure. What gap is that?
- Reviewers rank two draft notices; those rankings later teach a preference model that nudges the generator toward the preferred style and safety. Which alignment path is that?
- A county can call an existing general LLM, adapt one with its own notices, or attempt to pretrain a new model from scratch on a tiny local pile. Which foundation-model path fits a small office?
- A timetable desk has four options: rewrite the prompt, retrieve tonight’s PDF into the prompt, fine-tune on old tickets, or pretrain a new LLM. How should they order those from lightest to heaviest?
- A clerk-bot must cite this morning’s notice, which did not exist when the LLM was trained. The system first finds relevant passages, then conditions generation on them. What is that pattern?
- Notices are encoded into vectors; a new question is encoded the same way; nearby chunks come back. What RAG loop is that?
- Sailing exceptions change twice a week. Should the desk keep fine-tuning the model on each new exception sheet, or update a retrievable store?
- The bot invents a “Tuesday ice clause” in perfect prose. What is that failure?
- A team wants an NVIDIA framework path to customize a foundation model with enterprise data, not a serving-only microservice. Which tool fits?
- The customized model is accurate but too slow on the GPU they already have. Which official tool optimizes LLM inference?
- One group wants a packaged, portable microservice for a standard model; another already serves several custom models and wants a dedicated inference server. Which NVIDIA pair matches those jobs?
- A harbor authority lists three needs in order: adapt a foundation model to local notices, then speed up token generation, then put the result behind a live endpoint. Which NVIDIA sequence is that?
Software Development · 72 questions
- A city parks clerk pastes a trail-closure notice into a local Hugging Face script, and the model rejects the raw Unicode string. What must the script produce before a forward pass?
- A water-meter office sends three inspection notes of different lengths in one GPU call so the card is not woken three times. How does that batch become one rectangular tensor?
- A ferry-terminal kiosk needs a closed set of delay labels on incoming radio logs, not a drafted paragraph. Which architecture should the app load?
- A municipal archives desk computes a numeric vector for each new ordinance summary and stores that vector for later lookup. What role does that embedding play in the app?
- A fire-watch radio hut is short on GPU memory, and a colleague suggests serving the same refined weights at half precision. What kind of choice is that?
- A public-works help desk concatenates a year of pothole notes into one prompt, and the local model errors on input size. What constraint did the client ignore?
- A city animal-shelter clerk has two hundred labeled intake notes. A volunteer starts a random-weight network to be original. What transfer-learning first step should the script take?
- A municipal compost site wants food, yard, or contamination labels on hauler emails. How should transfer learning be wired in the script?
- A beach-permit booth's pretrained sentiment head already sorts compliment versus complaint mail well enough for Saturday traffic. A researcher wants a multi-week PEFT study before launch. What should the associate do?
- A historic-district newsletter classifier trained on open-web reviews keeps treating lot as a parking space when staff mean a parcel. How should the app fix that mismatch?
- A rec-center swim desk has a GPU that overheats when every layer is updated on forty labeled shift notes. Which associate transfer-learning knob should the script use?
- A city tree-nursery shop can either load a small encoder on the workshop GPU or POST text to a hosted NIM-style chat endpoint. When should the project stay on-box?
- A town-band festival intern loads model weights from checkpoint A and a tokenizer from checkpoint B because both listings look similar. Token IDs come back garbled. What framework rule was broken?
- A city food-pantry inventory script loads a transformer and then crawls, even though the machine has a GPU the process never used. What is the framework-project fix?
- A municipal bike-share desk is about to hand-parse a public CSV of repair notes for a Hugging Face project. What is the associate way to pull a named split?
- A city pond-census intern fine-tunes a small head on Friday and cannot find the weights on Monday when Triton should serve them. What handoff did the project miss?
- A city weather-kiosk helper mixes regex, a forward pass, and email sending in one long function. How should that framework project be split?
- A civic-center kiln desk has a refined classifier, and a colleague opens NVIDIA NeMo to serve it. Where should the finished artifact actually run?
- A city-orchard gleaning desk copies weights into a random home directory and Triton reports the model missing. What deploy unit does Triton expect?
- A community-band library app is a short Python script; the ticketing service is another backend that already speaks gRPC. Both need the same Triton-served encoder. How should each client connect?
- A municipal water-billing desk posts each meter note to Triton as its own request, and the GPU stays underused between arrivals. Which Triton feature lets nearby requests share one forward pass so throughput rises?
- After a parks-and-rec field-reservation helper raised Triton’s max queue delay so batches could grow, residents say replies feel stickier. What does that queue delay control?
- A building-permits counter needs a text classifier and a small summarizer live on the same GPU workstation. What is the associate way to host both?
- A transit lost-and-found page starts posting inferences the instant the Triton container process appears, and the first wave of lookups fail while weights still load. What should the client wait for?
- A snow-route dispatch desk exported the same refined encoder as a PyTorch checkpoint, an ONNX graph, and a TensorRT engine. How should they choose the Triton backend?
- A housing-inspection counter places a refined classifier into the live Triton repository, but the lobby kiosk still sends the old tensor names. How do they roll forward without breaking that kiosk?
- A poll-worker help page piles up while a volunteer keeps a laptop predict.py loop open and pastes questions by hand. Why move the model onto Triton?
- A tree-trimming request app tokenizes on the phone and sends IDs to Triton; another team wants the server to accept raw street notes. What is true about those two designs?
- A 311 call-center Triton config sets max batch size to an enormous number for speed, and a storm-day spike exhausts GPU memory. What is max batch size actually for?
- A marriage-license counter needs names and dates pulled from clerk notes before the desk closes today. What is the fastest associate Hugging Face entry point?
- A city-archives labeling script imports a hard-coded BertForSequenceClassification class, then the Hub checkpoint switches to another encoder family. Which loaders keep the same few lines working?
- An animal-shelter intake desk downloads the largest LLM on the Hub to label vaccination notes as complete or missing. What should they check first so the transformers API can run that job?
- A zoning-counter helper must return one closed land-use label for a short parcel note, not a generated paragraph. Which transformers call matches that head?
- A wastewater-lab intern loads a public notes split with Hugging Face datasets and maps a tokenizer over it before a small head update. What pairing is that?
- A cemetery-plot office helper must demo on a staff laptop with no GPU this weekend, then later call the county NIM endpoint. How should the app be structured?
- A tax-assessor chat helper concatenates User: lines by hand, and the instruct model replies like a raw completion. What should the code apply?
- A harbor-master log helper works with a Hub pipeline for a handful of clerks, then stalls when fifty concurrent lookups arrive. What is the associate next step?
- A public-library hold desk receives short notes that must be labeled hold-ready, fine-dispute, or room-booking. Which solution fits?
- A public-health clinic paste-bin of signup lines is free text, and the booking form needs name, date, and site fields. Which task fills those structured slots?
- A city-council minutes desk has five known speechwriters and a stack of unlabeled drafts. How should author attribution be implemented?
- A town-hall lobby plaque is 180 words about the 1912 clock tower. A visitor asks when the tower was dedicated, and the desk wants that year taken from the plaque rather than guessed. What kind of question answering should the app run?
- A public-works yard tags every field note as close, reopen, or inspect, and a vendor quotes a 70-billion-parameter chat model for that three-label job. What should the software team pick?
- A little-league registration desk runs a token classifier and then prints raw B-PER and I-PER strings on the parent receipt. What should the app do before the UI?
- A parks-permit office first needs the document type, then the applicant names, then an answer to a clerk question from that same text. How should the app compose those official NLP tasks?
- A historic-district intern scores an extractive QA helper with ordinary classification accuracy and reports the model broken. What smoke-check should the developer use?
- A veterans-memorial kiosk sends raw English bytes into a TensorRT-LLM-optimized engine that expects token IDs, then prints integer IDs to the visitor. What is missing from the live pipeline?
- A 311 chat window needs the first sentence quickly, while a nightly job must classify twelve thousand work-order lines. How should serving be tuned?
- A historical-society desk is handed a running NVIDIA NIM container and a base URL. How should the volunteer call the model?
- The permitting backend already speaks gRPC, while the public web form already calls the same live model over REST. What should the service-side caller do?
- A flood-watch office waits eight seconds on a blank box while a long briefing generates. What should the interactive path do?
- A city-council minutes chatbot appends every past turn until the endpoint returns a length error. Where is the fix?
- A public-works work-order summarizer is accurate in Hugging Face generate but misses the city's latency target on the same GPU. What is the associate implementation sequence?
- A community-college tutoring app already has working OpenAI-style client code and must stay on-prem. How should it call a local model?
- A cemetery-plot weekend import must embed two thousand short blurbs, while the visitor kiosk has one guest. How should batching be set?
- A trail-head weather helper freezes when the serving box is busy. What must the client add?
- A youth-makerspace copies a TensorRT-LLM engine built for the county H100 onto a classroom laptop GPU and it will not load. Why, and what should they keep?
- The regional transit HQ H100 is memory-tight on a larger decoder. Official additional material names TensorRT-LLM converting weights toward FP8 on H100. What should the team choose?
- A 4-H club helper needs a system preamble, two tools, and short memory. A volunteer wants twenty separate HTTP posts. What should the team pick?
- A community-theater playbill job embeds this week's show name in scattered string concatenations, and one file drops the no-ticket-prices rule. What should the code use?
- A planetarium lobby kiosk stores the whole conversation as one growing paragraph and the model loses who said what. How should chat memory be kept?
- A city parks desk pastes a weather-service bulletin into a helper that must return a short trail-closure checklist. Which application sequence should the developer implement?
- A rec-center membership desk needs a helper that greets a visitor, asks one clarifying question, then drafts a day-pass note. What should the application code do?
- A river-ferry helper should call a tide-table tool only when a rider names a dock and a departure window; otherwise it should answer from standing rules. Which application pattern fits that branch?
- A civic-garden hotline already has a plot-rules index built by another team. The chatbot must fetch the nearest three blurbs for watering on a holiday. What should the application code call?
- A food-pantry hours bot already received three similar hits from its vector client. A volunteer prints those hits to the terminal and never calls the model. What should the application do next?
- A community-pool bot should look up tonight's lap-lane hours from a small internal function when a swimmer asks when the lanes open. How should the orchestrator handle that?
- A town-clerk intern copies the same headers, retry loop, and message assembly into every feature branch for chatbot, summarizer, and extractor work. What should a reviewer ask for instead?
- A volunteer-fire board wants a weekend demo of newsletter Q&A and has nobody to own a model repository. How should the team kickstart?
- A school-cafeteria menu helper works on one intern laptop GPU and stalls when a class lines up. What should the team add before public lunch hours?
- A transit-shelter kiosk swaps hub checkpoints daily and nobody can say why answers drifted. What should the kickoff pin first?
- A city-minutes comment box accepts a whole novel and twenty browsers hit generate at once, and the GPU box falls over. Which software limits should the app set?
- A watershed-district office shipped this path: a local Hugging Face prototype, then an optional TensorRT-LLM optimize for a later decoder, then Triton or NIM in front, with LangChain only on the chatbot feature. Which reading of that stack is correct?
Experimentation · 66 questions
- A city archives desk wants a language model to draft finding-aid blurbs for newly accessioned boxes. A volunteer is already launching overnight adaptation jobs just to see what happens. What must be written down before the first run?
- A volunteer firehouse tags incident run reports. One overnight job changes the instruction set, the learning rate, and the base checkpoint together, and the crew cannot say which change helped. What is the design fault?
- A county floodplain desk tunes on last spring’s permit notes, peeks at a reserved autumn set after every epoch to pick a winner, and then reports that autumn score as the official result. Why is that score no longer a valid unseen test?
- A community-theater desk claims a customized model writes better playbill synopses but never scored the unmodified pretrained checkpoint on the same gold synopses. What is missing from the experiment?
- A municipal bike-share desk cannot replay last Tuesday’s best adaptation: two new tickets were added to the spreadsheet, the framework was upgraded, and nobody wrote down the random seed. What should have been frozen for a replayable run?
- A public-pool desk has a dozen adaptation attempts sitting in chat screenshots and cannot match which learning rate produced which score. What should they keep instead?
- A farmers-market inspection desk watches training loss fall for many extra epochs while the validation score has already peaked and begun to worsen. What should stop the run?
- A town clerk has one GPU evening and a wish list: full-weight training of a huge model, a small PEFT run, and a prompt-only trial. How should the evening be designed?
- A historical-society minute-book project prints a final number but never checks whether validation tracked training or diverged after epoch two. What should they read as the primary experiment instrument?
- A parks-permit desk declares model B the winner, but A was scored on last year’s questions with greedy decoding and B on this year’s questions with a looser sampler. What is wrong with that comparison?
- A food-bank intake desk has a few hundred labeled comments. One volunteer wants to train a transformer from scratch so it will know pantries; another wants to start from a general pretrained checkpoint and measure the difference on the same split. What is the fair experiment?
- A building-permit desk reuses a general encoder for work-order tags and cannot decide whether to keep the backbone frozen and train only a small head, or to unfreeze later layers. How should they decide?
- A trail-maintenance desk can start from a checkpoint trained on outdoor-incident notes or from one trained on restaurant reviews. Both are called pretrained. Which source is the better first transfer experiment, and why?
- A drawbridge office adapts a general model to bridge-status phrasing and then finds it can no longer answer the everyday weather questions it used to handle. What experimental metric is missing?
- A zoo script desk keeps adding labeled exhibit notes and wonders when training from scratch might catch a transferred model. How should they find out?
- A public-library bindery already agreed to start from a pretrained checkpoint. The remaining argument is whether to update a small PEFT adapter or the full weights on a modest GPU night. How should that argument be settled?
- A county-fair volunteer-crew worked-with graph grew from a classroom clique to tens of thousands of edges. The team wants the same shortest-path and importance queries they already trust. How should they design the runtime experiment?
- A bike-share desk wants to know whether a new who-rebalanced-whom graph recipe still surfaces the same key hubs as last season. What should be the experimental outcome?
- A watershed desk can feed an LLM either a flat list of sensor IDs or a short description of which sensors share a parent tributary — a graph-derived feature. They want to know which input helps a downstream summary. How should they compare those inputs?
- A parking-garage clerk has independent daily occupancy totals and no relationships among the rows. A consultant insists on cuGraph because NVIDIA. What should the clerk do?
- A licensing desk has a small labeled set of permit-status notes. One camp wants a compact encoder classifier; the other wants a decoder that writes the status in prose. How should they compare the two families?
- A parks desk extracts trail-hazard mentions from ranger notes. Staff can freeze the pretrained transformer and train a light head, or they can update more of the stack. What should that experiment measure?
- A planning packet is long, and a short token limit quietly drops the rezoning clause the label depends on. What factor belongs in the experiment, not in hidden preprocessing?
- A 311 desk can paste three graded tickets into the prompt or spend a night on a small supervised adaptation. How should they compare those two treatments?
- A clerk's office loves its note-tagging checkpoint and wants to reuse it unchanged to write inspection summaries. What should the experiment expect?
- A fire-inspection desk needs token tags for code faults. Full-weight training will not fit tonight's GPU; a parameter-efficient adapter will. What comparison should they run?
- A school-lunch allergen classifier won on 120 notes after one lucky shuffle. What should happen before the kitchen changes process?
- A transit delay model looks strong on last April's depot notes and fails on this January's ice-storm notes from another yard. What eval belongs in the reported experiment?
- A records clerk must point at the filing date already written in a short passage. One score requires the span to match gold characters exactly; another gives partial credit when the words mostly overlap. How should they pick?
- A water desk scores extractive 'where is valve 9?' answers with a translation-style n-gram overlap number and crowns a chatty restatement. What is wrong with that primary metric?
- A rec-center QA set accidentally includes the same 'what is the pool fee?' items that were pasted as few-shot hints in the prompt. Why is the comparison contaminated?
- A public-works office wants to know whether this week's checkpoint beats last month's on 'which pump tripped?' What makes that head-to-head valid?
- A parcel-lookup helper has a high automatic QA score, but clerks say it names the wrong lot even when a few words overlap the gold span. What should the experiment add?
- A complaint archive uses a second language model to grade long answers about who filed which case. The judge prefers florid answers that match its own style. What is that judge good for at associate depth?
- A bridge-closure question needs a two-sentence explanation of why the span closed, not a three-word date. How should the models be ranked?
- A permit-desk QA bake-off used sampling; model A won by two questions on a 40-item set. What should happen before they retire model B?
- A street-sign project tags asset names. Most tokens are ordinary words, so a model that marks everything other looks accurate. Which score should they trust?
- A housing desk classifies emergency versus routine notes. Routine notes dominate, and a model that always predicts routine looks accurate while missing every emergency. Which metric surfaces the rare class?
- A library-event bulletin summarizer wins a reference-overlap contest by copying sentences, while readers prefer a shorter rewrite that a human panel ranks higher. What do those two scores mean?
- A flood desk brags that an adapted model has lower perplexity on river chatter, then ships it as a flood / no-flood classifier without scoring those labels. What did they mix up?
- A city clerk must recast each parking-appeal ruling into a statute-mandated one-sentence formula that already has a gold restatement. Which automatic metric is a reasonable primary score for that constrained rewrite, yet a weak primary for open-ended chat?
- A public-library help desk lets one intern star thirty catalog-note summaries and calls the stack human evaluation. What must a real human-eval experiment include?
- A county elections office pulls one pretrained LLM and wants four jobs: token tags on polling-place names, a label for each precinct note, a short recap of a canvass memo, and a which-site question. How should they score the run?
- A municipal water desk's automatic scores say a new summarizer is better; night operators say it drops boil-water alerts. What should the experiment do?
- A city parks reservation desk has two standing instruction sheets for the same LangChain job. How should they compare those sheets?
- A public-works snow-route desk keeps the LangChain document-analysis path identical and only changes which LLM sits behind it. What is the experimental factor?
- A municipal archives desk can stuff a week of council minutes into one call, or map over each day's minutes and then reduce. How should they treat those two shapes?
- A transit lost-and-found desk tries the chain in a chat window whenever someone feels like it. What turns that habit into an experiment?
- A building-permits counter improved after someone quietly changed the template and the model name in the same afternoon. What must a scored chain experiment record?
- A recycling-center desk rewrites LangChain calls into cleaner objects, keeps the model and prompts the same, and wants to publish a quality win. How should that change be labeled?
- A public-housing waitlist office finds two Hub models tagged for the same kind of note classification. What is the right comparison?
- A zoning-board intern grabs the most-downloaded Hub checkpoint even though the model card describes a different language and task than their gazetteer notes. What should decide eligibility for the run?
- A tax-assessor team fine-tunes on a Hub dataset whose card warns that train and test share near-duplicate parcels. What should they do about the published test split?
- A fire-inspection loft wants to experiment by putting each candidate checkpoint behind a live server after every epoch. Where does the comparison belong?
- A municipal print-shop lab finishes a parameter-efficient run and tries to check in a full copy of the base model as the result. What should the run record as output?
- A flood-alert desk says they used the Hub summarizer last month; the default revision moved and this month's rerun does not match. What must a versioned run log include?
- A public-health inspection desk has a mid-size instruction set and one workstation GPU. Full-weight supervised fine-tuning would update every parameter; LoRA would freeze the base and train low-rank adapters. Which experiment should they run first, and why?
- A community-garden seed library raises adapter rank so the model can learn more patterns on a few dozen instruction cards. Training loss collapses while validation falls apart. How should rank be treated?
- A courthouse records office wants to adapt a larger pretrained model that will not fit in memory for ordinary LoRA. Which associate-depth choice lets the experiment run?
- A school-cafeteria menu desk needs one base model that can tag allergen notes or draft tray labels without training a second full copy. How can a multi-task experiment do that?
- A city band-shell programming desk has very little labeled program-note text and can either learn soft prompt or prefix parameters or attach a LoRA-style weight adapter. How should those two PEFT families be framed?
- A voter-registration desk can scrape two thousand noisy instruction pairs or carefully edit two hundred. A consultant says large language models just want scale. How should they design the supervised-adaptation data experiment?
- A historical-society desk writes eighteen precious question-and-answer cards, trains until the model recites them, then fails on a rephrased nineteenth card. What experimental response fits this instruction-tuning overfit?
- A municipal millpond office copies a pretraining-scale learning rate and a long epoch count into a LoRA run on a short instruction list. Validation dies by dinner. What should change?
- A city-council clerk wants replies that stay helpful and on-register. They can supervised-fine-tune on gold replies, run a human-feedback alignment experiment, or use SteerLM-style attribute control at inference. How should they choose?
- A parking-meter office must decide tonight's work: rewrite the standing instructions, run a LoRA-style adaptation on a small gold set, or retrieve the latest rate tables at ask-time. Facts change weekly; tone should sound like the traffic commissioner. Which experiment, or order of experiments, should they run, and which stack customizes a model?
Data Analysis and Visualization · 42 questions
- A city parks desk has a 180-row picnic-table repair log that already opens in a desktop spreadsheet. A vendor says RAPIDS is required before anyone can chart repair hours versus lumber cost. When is GPU-accelerated table work justified?
- A county clerk wants one analysis pass over marriage-license filings, courtroom calendars, and parking-ticket ledgers that together will not fit in a laptop session. Which official NVIDIA stack loads, joins, and summarizes those tables on GPU?
- A ferry-terminal authority has a decade of boarding tickets and a decade of tide-gauge readings. Staff export slices to a workstation and the joins time out. How should they integrate the two ledgers?
- A public-works yard filters pothole reports on the GPU, copies the result back to host memory to be safe, then sends it to the GPU again for a group-by. What is the problem with that bounce?
- A city-archives desk wants page-count spread, missing box labels, and how many lots came from each branch before anyone fine-tunes a model. A junior treats those plots as the experiment. What are the plots?
- A community-garden inventory exports a wide parquet of daily soil moisture and plot IDs. The analyst is about to open it in pandas and then move it to the GPU. How should that table land on a single GPU?
- A river-gauge station's 15-year stage archive is larger than the memory of the one GPU on the bench. How should they ingest it without discarding the science?
- A landfill scale-house now has two GPUs on the same box and a truck-ticket archive that barely fit on one device last season. How should they match the reader to scale?
- A school-bus depot is handed ridership files, delay logs, and weather extracts that use different date columns. Staff want to concatenate first and clean later. What is the right ingest order?
- A public-library intern writes a Python loop that walks every checkout row to recode a material label. The table is columnar and already on the GPU. What should they use instead?
- A municipal animal-shelter log has mixed string and number columns, blank feeding times, and volunteer rows that should be dropped. How should that prep stay on the GPU?
- A city streetlight shop wants hour-of-day, day-of-week, and a one-hot fixture code from a GPU table of work orders before any estimator runs. How should those columns be built?
- A planning-department intern asks why cuDF tables can be handed to other RAPIDS steps without a CSV round-trip. What keeps those GPU frames cheap to share?
- A city-hall records desk is building an LLM help corpus from ordinance PDFs. Before anyone trains, they need a column of token counts per page so they can drop empty pages and flag oversized ones. What is token length in that step?
- A wastewater-plant desk wants cuML clustering on temperature, turbidity, and chlorine. Raw units differ by orders of magnitude. What must happen before that GPU estimator runs?
- A bridge-inspection office shuffles report IDs into train and eval files but copies any row that mentions urgent into both so the eval looks realistic. What defect is that?
- A municipal pool desk wants a classifier for too-cold, too-hot, and just-right water notes. Almost every labeled slip says just right. What should they do first?
- A county-museum chatbot corpus has a pile of one-word captions and a few entire catalog PDFs pasted as one document. What plot helps them cap, split, or drop before ingest?
- A farmers-market inspector's scorecards omit the stall name on rainy Saturdays. A missingness plot shows that pattern. How should the desk read it?
- A community-theater costume shop has plenty of costume-ripped notes and almost no wig-melted notes. Staff propose paraphrasing the rare notes with a human check. What is that step?
- A municipal tree-nursery wants other desks to reuse their labeled planting notes. What short document records who collected the rows, the license, what the labels mean, and what the set is not for?
- A courthouse FAQ set was split by filename. Many eval questions are the train questions with a comma moved, and scores look perfect. What is the issue?
- A lighthouse-archive scrape keeps navigation chrome, click-here lines, and empty pages. What preprocessing step comes before any length plot or index build?
- A city recycling-sort yard already has a cleaned cuDF table of bin weights and contamination flags. Staff want scikit-learn-style fit and predict without leaving the GPU. Which official library is that?
- A civic-auditorium lost-and-found encodes item notes and wants to look at groups before anyone labels them. How should they explore that structure on the GPU?
- A municipal irrigation desk has labeled dry, seeping, and flooded sensor rows already on the GPU. How should they apply a supervised GPU estimator?
- A coastal-park ranger desk has a multi-GPU box and a table of trail-use, rainfall, and campsite-occupancy features that no longer fit on one device. How should they train boosted trees at that scale?
- A city marching-band warehouse intern offers to write a CUDA kernel to train boosted trees on the GPU table. What should they use instead?
- After a cuML cluster step, one cluster is almost all garbled OCR lines from a city tax-assessor's scanned invoices. How should that cluster be used?
- A city scooter-share desk has a huge two-column file of from-station and to-station trips. How do they get that edge table onto the GPU for analysis?
- A city parade-route planner's visitor-path graph has tens of millions of edges. NetworkX on a laptop will not finish overnight. Where do massive-scale graph analytics belong?
- After a cuGraph importance pass, a city food-bank warehouse wants those scores as extra columns on the original product table for later exploratory analysis. What should they do?
- A civic-arboretum FAQ is being loaded into a vector store. One pass cuts every 2,000 characters with no overlap, and questions that straddle a cut come back empty. Which ingest knobs matter?
- After encoding trail-marker plaques at a city nature preserve, a 2D sketch of the vectors shows hours and closing time in one clump and OCR garbage in another. What is that sketch?
- A city water-plant safety index was built with encoder A. A later desk embeds new PDFs with encoder B and appends them, and neighbors become nonsense. What must ingest record?
- A civic-orchestra handbook has numbered sections, but an ingest job still slices every N tokens, often in mid-sentence. How should those chunks be cut?
- A city aquarium desk is told to pick any embedder off the internet for a new visitor-FAQ index. Which official NVIDIA path is the associate retrieval and embedding stack?
- A county-fair transcript dump contains identical copies, copies with a typo, and paraphrased repeats of the same safety briefing. How does NeMo Data Curator distinguish those at selection depth?
- After dedup, a county-fair dump still has two-word fragments, other-language ads, and highly repetitive spam. What should happen before anyone builds a vector index?
- A rec-center desk's chemical, hours, and class-signup labels disagree on the same sentence, and named-entity spans for locker numbers sometimes include the next word. What data work comes first?
- A municipal print-shop wants author attribution on press notes and a small extractive question-answering set over equipment manuals. One author's notes are three words; another writes pages. What should they plot first?
- A lock-and-dam inspection office needs load, split, embed, and write-to-store as a repeatable pipeline, with a branch when a file fails to parse. Which official tools orchestrate that as data work?
Trustworthy AI · 30 questions
- A county-fair lost-and-found booth wants a flyer that calls their new FAQ bot “always correct.” What does a trustworthy-AI approach require instead?
- A public-library hold-notice helper drafts pickup messages that mention patron names. A vendor says privacy is “whatever the logging flag is set to.” How should the desk treat privacy?
- A community-garden tool-shed chatbot is meant to answer hours and key-checkout rules. Staff discover it will also draft ways to silence the shed alarm if asked. What trustworthy-AI gap is that?
- A farmers-market stall-draw assistant “just knows” which vendor gets the corner booth. Board members cannot tell what records it saw or why a stall was flagged. What principle must they require?
- A community-center yoga waitlist helper keeps offering prime class times to one neighborhood’s phrasing and almost none to another. Before any mitigation recipe, what principle is at issue?
- A volunteer fire-watch lookout ships a smoke-note generator with no note that it does not see live weather and must not be used to dispatch crews. What must they publish for operators?
- A town brass-band legally holds rehearsal tapes in its archive. Players never agreed those voices would train a public fan-singing bot. What distinction should the board make?
- A transit-pass office pastes a rider’s national ID and home address into a public LLM box to “make the replacement-card letter nicer.” How should that be treated?
- The same transit-pass office later finds those full prompts — IDs included — in a help-desk trace file shared with a vendor. The model’s letter looked fine. What failed?
- A quilt-guild workshop wants to fine-tune a helper on living makers’ quilt photos scraped from craft blogs. What consent check does trustworthy AI require?
- A letterpress shop bot emits several paragraphs that match a still-in-copyright binding manual. A marketer wants to publish them as original store copy. How should the shop treat that output?
- Four county animal-shelter clinics want a stronger intake-note classifier but will not ship charts to one shared server. Which NVIDIA-aware approach keeps records on each private site?
- A harvest-fair volunteer helper was trained mostly on notes from one township. Advice for other townships is curt or wrong. What bias source is that?
- Volunteer taggers at a folk-dance archive mark dialect speech as “rude” more often than the same content in the house dialect. The model then copies that habit. Where did the bias enter?
- A community-supported-agriculture FAQ bot treats one region’s produce names as “real” and rewrites others as errors. What kind of bias is that?
- A rec-center field-permit helper reports 94% “correct intent.” Almost all failures are names from one language community. What measurement step is required?
- After that subgroup gap appears, a vendor offers to stop reporting the missed community’s slice. What should the rec center do instead?
- A town-green concert ticket kiosk begins returning slurs when teenagers bait it after hours. How should the desk treat those outputs?
- A public-workshop tool-library fine-tunes a public LLM to write helpful equipment manuals. The same weights later get used to generate harmful bypass instructions for civic equipment. What risk is that?
- A dam-spillway office wants the bot to text evacuation orders with no person in the loop. What does trustworthy deployment require?
- A community-orchard kiosk invents a cultivar name and a made-up spray warning in a confident tone. Visitors would act on it. How should the orchard treat that output?
- A smokehouse visitor kiosk never says the bot cannot diagnose food-borne illness. A reviewer asks what visitors were told. What must the site provide?
- A city aviary chat should stay on hours, exhibits, and tickets. The desk is about to add a stern system prompt and hope. Which NVIDIA trust control should they select instead?
- The aviary bot starts giving stock-market tips when asked. What kind of rail keeps it on its subject?
- The same aviary bot will quote anonymous forum posts that include slurs if retrieval is left open. What rail limits harmful language and untrusted sources in replies?
- A later version of the aviary app can call a hatch-room door-schedule API. Someone tries to make the LLM issue an unlock. What rail is required?
- A county behavioral-health desk must run inference on notes that cannot be visible to the cloud operator while the job is in memory. Which hardware trust feature matches that need?
- A township purchasing clerk will not accept a mystery checkpoint for a notice-writing model. What NVIDIA documentation format should the vendor supply?
- A surveyor’s plat-book assistant answers property-line questions with no footnote. What trust practice should the office require?
- A cider-orchard co-op wants to “just ship the bot Friday” with only a system prompt. Domain 5’s kickoff skill is a safe, effective, and scalable generative solution. What must they put in place together?
These questions are original practice material and are NOT actual exam questions or brain-dump content. All vendor marks are trademarks of their respective owners. This site is not affiliated with, endorsed by, or sponsored by the exam vendor.