[Core][MRV2] Support eagle3 spec decode with pipeline parallel (#50514)
Signed-off-by: Yongqin Wang <yongqinwang@roblox.com> Co-authored-by: Cursor <noreply@cursor.com> Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>
This commit is contained in:
co-authored by
Cursor
Wentao Ye
parent
d4ef5e53fc
commit
d87a440f88
@@ -245,9 +245,7 @@ class DFlashLagunaForCausalLM(nn.Module, SupportsEagle3):
|
||||
raise ValueError("Laguna DFlash config requires `draft_vocab_size`.")
|
||||
self.has_own_embed_tokens = False
|
||||
self.has_own_lm_head = False
|
||||
target_layer_num = vllm_config.model_config.get_num_layers(
|
||||
vllm_config.parallel_config
|
||||
)
|
||||
target_layer_num = vllm_config.model_config.get_total_num_hidden_layers()
|
||||
self.config.target_layer_count = target_layer_num
|
||||
target_vocab_size = vllm_config.model_config.get_vocab_size()
|
||||
if self.config.draft_vocab_size != target_vocab_size:
|
||||
|
||||
Reference in New Issue
Block a user