Unverified Commit d3be9710 authored by Stefan He's avatar Stefan He Committed by GitHub
Browse files

correct the tp_plan logic (#8850)

parent 3e7ff1ab
......@@ -211,16 +211,13 @@ class TransformersForCausalLM(nn.Module):
Apply the model's tensor parallelization plan.
Currently only supports linear layers.
"""
if not self.model.supports_tp_plan:
if tp_size <= 1:
return
tp_plan = getattr(self.model.config, "base_model_tp_plan", None) or {}
if not tp_plan and self.tp_size > 1:
raise ValueError(
f"{type(self.model)} does not support tensor parallel yet!"
)
tp_plan = getattr(self.model.config, "base_model_tp_plan", None) or {}
def _tensor_parallel(module: nn.Module, prefix: str = ""):
for child_name, child_module in module.named_children():
qual_name = maybe_prefix(prefix, child_name)
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment