[6/N] pass whole config to inner model (#10205)

Signed-off-by: youkaichao <youkaichao@gmail.com>

vllm/model_executor/models/olmo.py

@@ -46,7 +46,8 @@ from vllm.sequence import IntermediateTensors
 
 from .interfaces import SupportsPP
 from .utils import (is_pp_missing_parameter,
-                    make_empty_intermediate_tensors_factory, make_layers)
+                    make_empty_intermediate_tensors_factory, make_layers,
+                    maybe_prefix)
 
 
 class OlmoAttention(nn.Module):
@@ -224,12 +225,13 @@ class OlmoDecoderLayer(nn.Module):
 @support_torch_compile
 class OlmoModel(nn.Module):
 
-    def __init__(self,
-                 config: OlmoConfig,
-                 cache_config: Optional[CacheConfig] = None,
-                 quant_config: Optional[QuantizationConfig] = None,
-                 prefix: str = ""):
+    def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
         super().__init__()
+
+        config = vllm_config.model_config.hf_config
+        cache_config = vllm_config.cache_config
+        quant_config = vllm_config.quant_config
+
         self.config = config
 
         self.embed_tokens = VocabParallelEmbedding(config.vocab_size,
@@ -291,17 +293,13 @@ class OlmoForCausalLM(nn.Module, SupportsPP):
     Extremely barebones HF model wrapper.
     """
 
-    def __init__(
-        self,
-        vllm_config: VllmConfig,
-        prefix: str = "",
-    ) -> None:
+    def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
         super().__init__()
         config = vllm_config.model_config.hf_config
-        cache_config = vllm_config.cache_config
         quant_config = vllm_config.quant_config
         self.config = config
-        self.model = OlmoModel(config, cache_config, quant_config)
+        self.model = OlmoModel(vllm_config=vllm_config,
+                               prefix=maybe_prefix(prefix, "model"))
         if config.tie_word_embeddings:
             self.lm_head = self.model.embed_tokens
         else: