Neural architectures, PyTorch, Transformers, and RL.
Forward-backward propagation, loss landscapes, optimization algorithms, and neural networks.
4 Courses
Custom autograd functions, mixed-precision training (AMP), DistributedDataParallel (DDP), and FSDP.
3 Courses
Vision-language architectures, cross-attention projection, CLIP embeddings, and multimodal pipelines.
4 Courses
Score-based generative models, DDPM/DDIM samplers, latent diffusion architectures, and conditioning mechanisms.
3 Courses
Markov decision processes, policy gradients (PPO), value estimation, and reward model training.
4 Courses
Sparse routing algorithms, expert capacity factor, load balancing loss, and distributed expert parallelism.
3 Courses
Acoustic feature extraction, spectrograms, Whisper ASR architecture, and neural audio synthesis.
3 Courses
Hardware profiling, memory footprint reduction, gradient checkpointing, and FlashAttention.
3 Courses