Transformers adds FP8 backend for QuantizedCache
Native PyTorch FP8 concat replaces a defensive workaround, with layout fixes aimed at future kernels.
By tensor#kv-cache
Ollama ROCm on Strix Halo returns prior answers after context overflow
On gfx1151 hardware, truncated over-window prompts can replay another request's completion instead of generating a new one.
By tensor