Encode/decode utility
dead simple pip install torch that just works
High-level inference engine for UniLLM
Hybrid KV cache (RadixAttention + PagedAttention) for UniLLM
Core inference runtime for UniLLM with 47 model architectures
Request scheduling with continuous batching for UniLLM
AI-powered git commit message generator that delegates to terminal AI agents
Shared traits and types for zen* image codecs