mirror of
https://github.com/zml/zml.git
synced 2026-11-16 19:11:09 +00:00
Move manual regions out of flash attention kernels and use `manualComputation2` for every shared attention backend. Also fully replicate the embedding executable outputs in Qwen and Llama. Partial graph constraints leave their batch and sequence dimensions open, which is invalid at executable boundaries. Fix two further blockers exposed by the Qwen run that we forgot to move recently: softplus comparisons must use the input dtype and ROCm DNN runfiles must include the MIOpen kernel archives. |
History
|
|
|---|---|---|
| .. | ||
| benchmark | ||
| io | ||
| llm | ||
| mnist | ||
| sharding | ||