Rename fallback model and refactor supported models section (#15829)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>

vllm/model_executor/models/registry.py

@@ -202,7 +202,7 @@ _SPECULATIVE_DECODING_MODELS = {
     "MLPSpeculatorPreTrainedModel": ("mlp_speculator", "MLPSpeculator"),
 }
 
-_FALLBACK_MODEL = {
+_TRANSFORMERS_MODELS = {
     "TransformersForCausalLM": ("transformers", "TransformersForCausalLM"),
 }
 # yapf: enable
@@ -213,7 +213,7 @@ _VLLM_MODELS = {
     **_CROSS_ENCODER_MODELS,
     **_MULTIMODAL_MODELS,
     **_SPECULATIVE_DECODING_MODELS,
-    **_FALLBACK_MODEL,
+    **_TRANSFORMERS_MODELS,
 }
 
 # This variable is used as the args for subprocess.run(). We
@@ -427,7 +427,7 @@ class _ModelRegistry:
         normalized_arch = list(
             filter(lambda model: model in self.models, architectures))
 
-        # make sure Transformers fallback are put at the last
+        # make sure Transformers backend is put at the last as a fallback
         if len(normalized_arch) != len(architectures):
             normalized_arch.append("TransformersForCausalLM")
         return normalized_arch

vllm/model_executor/models/transformers.py

@@ -401,7 +401,7 @@ class TransformersForCausalLM(nn.Module, SupportsQuant, SupportsLoRA,
         self.make_empty_intermediate_tensors = (
             self.model.make_empty_intermediate_tensors)
 
-    # FIXME(Isotr0py): Don't use any weights mapper for Transformers fallback,
+    # FIXME(Isotr0py): Don't use any weights mapper for Transformers backend,
     # this makes thing complicated. We need to remove this mapper after refactor
     # `TransformersModel` in the future.
     @property