[Bugfix][Model] Refactor OLMo model to support new HF format in trans…

…formers 4.40.0 (vllm-project#4324)

Co-authored-by: Woosuk Kwon <[email protected]>

README.md

@@ -74,7 +74,7 @@ vLLM seamlessly supports many Hugging Face models, including the following archi
 - Mistral (`mistralai/Mistral-7B-v0.1`, `mistralai/Mistral-7B-Instruct-v0.1`, etc.)
 - Mixtral (`mistralai/Mixtral-8x7B-v0.1`, `mistralai/Mixtral-8x7B-Instruct-v0.1`, `mistral-community/Mixtral-8x22B-v0.1`, etc.)
 - MPT (`mosaicml/mpt-7b`, `mosaicml/mpt-30b`, etc.)
-- OLMo (`allenai/OLMo-1B`, `allenai/OLMo-7B`, etc.)
+- OLMo (`allenai/OLMo-1B-hf`, `allenai/OLMo-7B-hf`, etc.)
 - OPT (`facebook/opt-66b`, `facebook/opt-iml-max-30b`, etc.)
 - Orion (`OrionStarAI/Orion-14B-Base`, `OrionStarAI/Orion-14B-Chat`, etc.)
 - Phi (`microsoft/phi-1_5`, `microsoft/phi-2`, etc.)

docs/source/models/supported_models.rst

@@ -101,7 +101,7 @@ Alongside each architecture, we include some popular models that use it.
     -
   * - :code:`OLMoForCausalLM`
     - OLMo
-    - :code:`allenai/OLMo-1B`, :code:`allenai/OLMo-7B`, etc.
+    - :code:`allenai/OLMo-1B-hf`, :code:`allenai/OLMo-7B-hf`, etc.
     -
   * - :code:`OPTForCausalLM`
     - OPT, OPT-IML

requirements-dev.txt

@@ -26,7 +26,6 @@ requests
 ray
 peft
 awscli
-ai2-olmo # required for OLMo
 
 # Benchmarking
 aiohttp

vllm/model_executor/models/__init__.py

@@ -42,7 +42,7 @@
     "MptForCausalLM": ("mpt", "MPTForCausalLM"),
     "MPTForCausalLM": ("mpt", "MPTForCausalLM"),
     "MiniCPMForCausalLM": ("minicpm", "MiniCPMForCausalLM"),
-    "OLMoForCausalLM": ("olmo", "OLMoForCausalLM"),
+    "OlmoForCausalLM": ("olmo", "OlmoForCausalLM"),
     "OPTForCausalLM": ("opt", "OPTForCausalLM"),
     "OrionForCausalLM": ("orion", "OrionForCausalLM"),
     "PhiForCausalLM": ("phi", "PhiForCausalLM"),