Results for “activation”
10 skills
orchestra-research Pyvene Interventions
Perform causal interventions on PyTorch models using pyvene's declarative framework for causal tracing, activation patching, and interchange intervention training.
10.4k · bundle
nvidia Nemo Mbridge Perf Cpu Offloading
Configure and validate CPU offloading for Megatron Bridge training, including activation offloading and optimizer state offloading with HybridDeviceOptimizer.
2.2k · bundle
nvidia Nemo Mbridge Perf Memory Tuning
Reduces peak GPU memory in Megatron Bridge training by applying expandable segments, parallelism resizing, activation recompute, and CPU offloading constraints.
2.2k · bundle
orchestra-research Moe Training
Train Mixture of Experts (MoE) models using DeepSpeed or HuggingFace, covering architectures, routing, load balancing, and expert parallelism.
10.4k · bundle
More results
orchestra-research Nnsight Remote Interpretability
Run interpretability experiments on neural network internals using nnsight, with optional NDIF remote execution for massive models.
10.4k · bundle
majiayu000 Awq Quantization
Quantize large language models to 4-bit precision using activation-aware weight quantization, reducing memory footprint and speeding up inference with minimal accuracy loss.
567 · bundle
orchestra-research Sparse Autoencoder Training
Train and analyze Sparse Autoencoders (SAEs) using SAELens to decompose neural network activations into interpretable features for mechanistic interpretability research.
10.4k · bundle
orchestra-research Awq Quantization
Quantize large language models to 4-bit using activation-aware weight quantization, achieving ~3x speedup with minimal accuracy loss for deployment on limited GPU memory.
10.4k · bundle
mukul975 Detecting Data And Model Poisoning
Detect poisoned training data and backdoored models across the ML pipeline using statistical analysis, activation clustering, and spectral signatures.
24.6k · bundle
lord1egypt Sparse Autoencoder Training
Trains and analyzes Sparse Autoencoders (SAEs) with SAELens to decompose neural network activations into interpretable features, covering loading pre-trained SAEs, training custom ones, and feature steering.
2