Dipika Sikka đ
@dsikka84
MLE @RedHat_AI @neuralmagic
I work on LLM Compressor & Speculators
17 Following 157 Followers
@RedHat_AI Some performance plots to demonstrate this, illustrating per-user interactivity combining our Quantized NVFP4 model with our trained DSpark speculator
Our LLM Compressor team has released NVFP4 and FP8 Kimi-K2.6 checkpoints - check them out!
