Add Lemur model (#2584)

Co-authored-by: Roberto Ugolotti <Roberto.UGOLOTTI@ec.europa.eu>

docs/model_support.md

@@ -48,6 +48,7 @@
 - [HuggingFaceH4/starchat-beta](https://huggingface.co/HuggingFaceH4/starchat-beta)
 - [HuggingFaceH4/zephyr-7b-alpha](https://huggingface.co/HuggingFaceH4/zephyr-7b-alpha)
 - [Xwin-LM/Xwin-LM-7B-V0.1](https://huggingface.co/Xwin-LM/Xwin-LM-70B-V0.1)
+- [OpenLemur/lemur-70b-chat-v1](https://huggingface.co/OpenLemur/lemur-70b-chat-v1)
 - Any [EleutherAI](https://huggingface.co/EleutherAI) pythia model such as [pythia-6.9b](https://huggingface.co/EleutherAI/pythia-6.9b)
 - Any [Peft](https://github.com/huggingface/peft) adapter trained on top of a
   model above.  To activate, must have `peft` in the model path.  Note: If

fastchat/conversation.py

@@ -639,6 +639,21 @@ def get_conv_template(name: str) -> Conversation:
     )
 )
 
+# Lemur-70b-chat default template
+# reference: https://huggingface.co/OpenLemur/lemur-70b-chat-v1#generation
+register_conv_template(
+    Conversation(
+        name="lemur-70b-chat",
+        system_template="""<|im_start|>system
+{system_message}""",
+        system_message="""You are a helpful, respectful, and honest assistant.""",
+        roles=("<|im_start|>user", "<|im_start|>assistant"),
+        sep_style=SeparatorStyle.CHATML,
+        sep="<|im_end|>",
+        stop_token_ids=[32002, 0],
+    )
+)
+
 # MPT-30b-instruct default template
 # reference: https://huggingface.co/mosaicml/mpt-30b-instruct#formatting
 register_conv_template(

fastchat/model/model_adapter.py

@@ -1681,6 +1681,18 @@ def get_default_conv_template(self, model_path: str) -> Conversation:
         return get_conv_template("vicuna_v1.1")
 
 
+class LemurAdapter(BaseModelAdapter):
+    """The model adapter for OpenLemur/lemur-70b-chat-v1"""
+
+    use_fast_tokenizer = False
+
+    def match(self, model_path: str):
+        return "lemur-70b-chat" in model_path.lower()
+
+    def get_default_conv_template(self, model_path: str) -> Conversation:
+        return get_conv_template("lemur-70b-chat")
+
+
 # Note: the registration order matters.
 # The one registered earlier has a higher matching priority.
 register_model_adapter(PeftModelAdapter)
@@ -1742,6 +1754,7 @@ def get_default_conv_template(self, model_path: str) -> Conversation:
 register_model_adapter(Llama2ChangAdapter)
 register_model_adapter(ZephyrAdapter)
 register_model_adapter(XwinLMAdapter)
+register_model_adapter(LemurAdapter)
 
 # After all adapters, try the default base adapter.
 register_model_adapter(BaseModelAdapter)

fastchat/model/model_registry.py

@@ -339,6 +339,13 @@ def get_model_info(name: str) -> ModelInfo:
     "Chat models developed by Xwin-LM team",
 )
 
+register_model_info(
+    ["lemur-70b-chat"],
+    "Lemur-Chat",
+    "https://huggingface.co/OpenLemur/lemur-70b-chat-v1",
+    "an openly accessible language model optimized for both natural language and coding capabilities ",
+)
+
 register_model_info(
     ["Mistral-7B-OpenOrca"],
     "Open-Orca",