Update README.md

README.md

@@ -38,7 +38,7 @@ LoRAX (LoRA eXchange) is a framework that allows users to serve over a hundred f
 - 🚅 **Dynamic Adapter Loading:** allows each set of fine-tuned LoRA weights to be loaded from storage just-in-time as requests come in at runtime, without blocking concurrent requests.
 - 🏋️‍♀️ **Tiered Weight Caching:** supports fast exchanging of LoRA adapters between requests, and offloading of adapter weights to CPU and disk to avoid out-of-memory errors.
 - 🧁 **Continuous Multi-Adapter Batching:** a fair scheduling policy for optimizing aggregate throughput of the system that extends the popular continuous batching strategy to work across multiple sets of LoRA adapters in parallel.
-- 👬 **Optimized Inference:**  high throughput and low latency optimizations including tensor parallelism, [continuous batching](https://github.com/predibase/lorax/tree/main/router) across different adapters, [flash-attention](https://arxiv.org/abs/2307.08691), [paged attention](https://arxiv.org/abs/2309.06180), [SGMV](https://arxiv.org/abs/2310.18547) quantization with [bitsandbytes](https://github.com/TimDettmers/bitsandbytes) and [GPT-Q](https://arxiv.org/abs/2210.17323), token streaming, weight prefetching and offloading.
+- 👬 **Optimized Inference:**  high throughput and low latency optimizations including tensor parallelism, [continuous batching](https://github.com/predibase/lorax/tree/main/router) across different adapters, pre-compiled CUDA kernels ([flash-attention](https://arxiv.org/abs/2307.08691), [paged attention](https://arxiv.org/abs/2309.06180), [SGMV](https://arxiv.org/abs/2310.18547)), quantization with [bitsandbytes](https://github.com/TimDettmers/bitsandbytes) and [GPT-Q](https://arxiv.org/abs/2210.17323), token streaming, weight prefetching and offloading.
 - 🚢  **Ready for Production** prebuilt Docker images, Helm charts for Kubernetes, Prometheus metrics, and distributed tracing with Open Telemetry.
 - 🤯 **Free for Commercial Use:** Apache 2.0 License. Enough said 😎.