biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Misc] Update Fused MoE weight loading (#7334)

Browse Source
This commit is contained in:
Dipika Sikka
2024-08-13 14:57:45 -04:00
committed by GitHub
parent fb377d7e74
commit d3bdfd3ab9
6 changed files with 269 additions and 206 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/model_executor/models/deepseek_v2.py
View File

@@ -593,7 +593,7 @@ class DeepseekV2ForCausalLM(nn.Module):
weight_loader = param.weight_loader
weight_loader(param,
loaded_weight,
weight_name,
name,
shard_id=shard_id,
expert_id=expert_id)
break

2
vllm/model_executor/models/jamba.py
View File

@@ -930,7 +930,7 @@ class JambaForCausalLM(nn.Module, HasInnerState):
weight_loader = param.weight_loader
weight_loader(param,
loaded_weight,
weight_name,
name,
shard_id=shard_id,
expert_id=expert_id)
break

2
vllm/model_executor/models/mixtral.py
View File

@@ -455,7 +455,7 @@ class MixtralForCausalLM(nn.Module, SupportsLoRA):
weight_loader = param.weight_loader
weight_loader(param,
loaded_weight,
weight_name,
name,
shard_id=shard_id,
expert_id=expert_id)
break

2
vllm/model_executor/models/qwen2_moe.py
View File

@@ -492,7 +492,7 @@ class Qwen2MoeForCausalLM(nn.Module):
weight_loader = param.weight_loader
weight_loader(param,
loaded_weight,
weight_name,
name,
shard_id=shard_id,
expert_id=expert_id)
break