Orchestra-Research

Provides PyTorch-native distributed LLM pretraining using torchtitan with 4D parallelism (FSDP2, TP, PP, CP).

Provides guidance for mechanistic interpretability research using TransformerLens to inspect and manipulate transformer internals via…

Fine-tune LLMs using reinforcement learning with TRL - SFT for instruction tuning, DPO for preference alignment, PPO/GRPO for reward…

Expert guidance for fast fine-tuning with Unsloth - 2-5x faster training, 50-80% less memory, LoRA/QLoRA optimization

Provides guidance for training LLMs with reinforcement learning using verl (Volcano Engine RL).

Serves LLMs with high throughput using vLLM's PagedAttention and continuous batching.

Track ML experiments with automatic logging, visualize training in real-time, optimize hyperparameters with sweeps, and manage model…

OpenAI's general-purpose speech recognition model.