minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

[Auto Sync] Update grok.py (20260213) (#18765)

Browse Source 
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Cheng Wan <54331508+ch-wan@users.noreply.github.com>
This commit is contained in:
Lianmin Zheng
2026-02-12 18:41:41 -08:00
committed by GitHub
co-authored by github-actions[bot] Cheng Wan
parent d5f66fec15
commit c56a5efbaa
1 changed files with 1 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/sglang/srt/models/grok.py
+1 -8
View File
@@ -690,19 +690,12 @@ class Grok1ForCausalLM(nn.Module):
config, "load_presharded_embedding", False config, "load_presharded_embedding", False
) )
self.is_weights_presharded = (
self.load_presharded_mlp
or self.load_presharded_moe
or self.load_presharded_attn
or self.load_presharded_embedding
)
default_replicate_lm_head = False default_replicate_lm_head = False
self.replicate_lm_head = getattr( self.replicate_lm_head = getattr(
config, "replicate_lm_head", default_replicate_lm_head config, "replicate_lm_head", default_replicate_lm_head
) )
if self.is_weights_presharded: if get_tensor_model_parallel_world_size() > 1:
setattr(DefaultModelLoader, "_prepare_weights", _prepare_presharded_weights) setattr(DefaultModelLoader, "_prepare_weights", _prepare_presharded_weights)
self.replicate_embedding = getattr(config, "replicate_embedding", False) self.replicate_embedding = getattr(config, "replicate_embedding", False)