NVIDIA Certified Associate - Multimodal Generative AI (NCA-MMGA) Flashcards¶
58 cards from this guide's fact sheet and notes. Click a question to reveal the answer.
To study with spaced repetition, download flashcards.csv and import it into Anki or Quizlet (comma separated, first row is the header). Back to the study guide.
Exam logistics¶
How long is the NVIDIA Certified Associate - Multimodal Generative AI (NCA-MMGA) exam?
60 minutes
How many questions are on the NVIDIA Certified Associate - Multimodal Generative AI (NCA-MMGA) exam?
50-60 questions
What is the passing score for NVIDIA Certified Associate - Multimodal Generative AI (NCA-MMGA)?
Not officially published
What does the NVIDIA Certified Associate - Multimodal Generative AI (NCA-MMGA) exam cost?
$135 USD
How long is NVIDIA Certified Associate - Multimodal Generative AI (NCA-MMGA) valid for?
2 years
Multimodal fundamentals¶
LLaVA (multimodal fundamentals)
Visual instruction tuning on LLaMA
NVIDIA VILA (multimodal fundamentals)
NVIDIA's vision-language model family
GPT-4V (multimodal fundamentals)
OpenAI's multimodal model
Gemini (multimodal fundamentals)
Google's multimodal model
Medical imaging (multimodal fundamentals)
Analyze X-rays and scans with text descriptions
Document understanding (multimodal fundamentals)
OCR + language understanding for forms
Retail (multimodal fundamentals)
Product image search and description
Quality control (multimodal fundamentals)
Visual inspection with natural language reports
Marketing (multimodal fundamentals)
Generate images from text descriptions
Design (multimodal fundamentals)
Create variations of visual concepts
Education (multimodal fundamentals)
Generate illustrations for learning materials
Entertainment (multimodal fundamentals)
Visual storytelling and concept art
Image description (multimodal fundamentals)
Generate alt text for visually impaired users
Speech-to-text (multimodal fundamentals)
Transcription for hearing impaired users
Text-to-speech (multimodal fundamentals)
Audio content for visually impaired users
Sign language (multimodal fundamentals)
Translation between sign and text/speech
Image video generation¶
VAE Encoder (image video generation)
Compresses image to latent space (smaller)
U-Net (image video generation)
Predicts noise in latent space
Text Encoder (image video generation)
Converts prompt to conditioning embeddings (CLIP)
VAE Decoder (image video generation)
Decompresses latent back to image
Seed (image video generation)
Random seed for reproducible generation
Resolution (image video generation)
Output image dimensions
Aspect ratio (image video generation)
Width-to-height ratio
CFG scale (image video generation)
Prompt adherence strength
Steps (image video generation)
Number of denoising steps
img2img (image video generation)
Transform existing image based on prompt
Autoregressive (image video generation)
generate frames sequentially
Latent video diffusion (image video generation)
3D latent space (spatial + temporal)
Frame interpolation (image video generation)
generate key frames, interpolate between
Text speech¶
Word Error Rate (WER) (text speech)
Primary quality metric
Real-time factor (text speech)
Processing time vs audio duration
Streaming vs batch (text speech)
Real-time transcription vs post-processing
Domain adaptation (text speech)
Custom vocabulary for specific industries
Naturalness (text speech)
How human-like the speech sounds
Prosody (text speech)
Rhythm, stress, and intonation patterns
Voice cloning (text speech)
Create synthetic voice from audio samples
Multi-speaker (text speech)
Single model with multiple voice options
Emotion control (text speech)
Adjust emotional tone of speech
Sample rate (text speech)
Number of audio samples per second
Channels (text speech)
Mono (1) or stereo (2)
Bit depth (text speech)
Dynamic range resolution
Duration (text speech)
Length of audio segment
Nvidia tools applications¶
📖 NIM Model Catalog (nvidia tools applications)
Browse available models
BLEU (nvidia tools applications)
n-gram overlap for translation/captioning
ROUGE (nvidia tools applications)
Recall-based metric for summarization
CIDEr (nvidia tools applications)
Consensus-based image description evaluation
BERTScore (nvidia tools applications)
Semantic similarity using BERT embeddings
Healthcare (nvidia tools applications)
Medical image analysis with text reports
Retail (nvidia tools applications)
Visual search and product description
Education (nvidia tools applications)
Multimodal content creation for learning
Manufacturing (nvidia tools applications)
Visual inspection with automated reporting
Media (nvidia tools applications)
Content generation for marketing and entertainment
Accessibility (nvidia tools applications)
Image description, speech-to-text, TTS