model : inkling2_loop share the loop slots after the scale pass (#32704)
The loop slots copied the physical layers at the end of load_arch_tensors, before the generic pass attaches the optional .scale tensors, so the later passes through the core ran NVFP4 experts without their per-expert scales. Add an optional post_arch_tensors hook that runs after the scale pass and share the slots there.
This commit is contained in:
1 parent
838d942438
commit
690b87d881
4 files changed
+7
No files matched your search
@@ -2019,6 +2019,7 @@ bool llama_model_base::load_tensors(llama_model_loader & ml) {
|
||||
output_in_s = create_tensor(tn(LLM_TENSOR_OUTPUT, "input_scale"), {1}, TENSOR_NOT_REQUIRED);
|
||||
}
|
||||
}
|
||||
post_arch_tensors(ml);
|
||||
}
|
||||
|
||||
ml.done_getting_tensors();
|
||||
|
||||
Reference in new issue
Block a user