Skip to content

Conversation

@llbdyiu66
Copy link
Collaborator

No description provided.

pre_alloc_memory: 60
tensorwise_offload_optimizer: true
use_fused_rms_norm: true
fuse_rms_norm: true
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

deepseek v3 pretrain修改需要和张博确认

self.freq_allocation = config.get("freq_allocation", 0)
if self.rope_3d:
assert not self.fuse_rope, "does not support fuse rope when rope_3d is on for now."
assert not self.apply_rope_fusion, "does not support fuse rope when rope_3d is on for now."
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

确认一下这里的修改对erniekit是否有影响

self.gate_up_proj_name = gate_up_proj_name

if self.fuse_up_gate:
self.fuse_swiglu = True
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

为什么这里默认为True

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants