[Misc] Add uninitialized params tracking for AutoWeightsLoader (#10327)

Signed-off-by: Isotr0py <2037008807@qq.com>

vllm/model_executor/models/jamba.py

@@ -1,5 +1,5 @@
 """Inference-only Jamba model."""
-from typing import Iterable, List, Optional, Tuple
+from typing import Iterable, List, Optional, Set, Tuple
 
 import torch
 from torch import nn
@@ -462,7 +462,8 @@ class JambaForCausalLM(nn.Module, HasInnerState, SupportsLoRA):
         next_tokens = self.sampler(logits, sampling_metadata)
         return next_tokens
 
-    def load_weights(self, weights: Iterable[Tuple[str, torch.Tensor]]):
+    def load_weights(self, weights: Iterable[Tuple[str,
+                                                   torch.Tensor]]) -> Set[str]:
         stacked_params_mapping = [
             # (param_name, shard_name, shard_id)
             ("qkv_proj", "q_proj", "q"),
@@ -479,6 +480,7 @@ class JambaForCausalLM(nn.Module, HasInnerState, SupportsLoRA):
             num_experts=self.config.num_experts)
 
         params_dict = dict(self.named_parameters())
+        loaded_params: Set[str] = set()
         for name, loaded_weight in weights:
             if "rotary_emb.inv_freq" in name:
                 continue
@@ -534,6 +536,8 @@ class JambaForCausalLM(nn.Module, HasInnerState, SupportsLoRA):
                     weight_loader = getattr(param, "weight_loader",
                                             default_weight_loader)
                     weight_loader(param, loaded_weight)
+            loaded_params.add(name)
+        return loaded_params
 
 
 def _is_moe_layer(name: str):