:zap: [Enhance] Add new model yi-1.5-34b, and remove unavailable models

apis/chat_api.py

@@ -64,7 +64,6 @@ class ChatAPIApp:
         raise INVALID_API_KEY_ERROR
 
     class ChatCompletionsPostItem(BaseModel):
-
         model: str = Field(
             default="nous-mixtral-8x7b",
             description="(str) `nous-mixtral-8x7b`",

constants/models.py

@@ -2,11 +2,12 @@ MODEL_MAP = {
     "mixtral-8x7b": "mistralai/Mixtral-8x7B-Instruct-v0.1",  # [Recommended]
     "nous-mixtral-8x7b": "NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO",
     "mistral-7b": "mistralai/Mistral-7B-Instruct-v0.2",
-    # "openchat-3.5": "openchat/openchat-3.5-0106",
+    "yi-1.5-34b": "01-ai/Yi-1.5-34B-Chat",
     "gemma-7b": "google/gemma-1.1-7b-it",
-    "command-r-plus": "CohereForAI/c4ai-command-r-plus",
-    "llama3-70b": "meta-llama/Meta-Llama-3-70B-Instruct",
-    "zephyr-141b": "HuggingFaceH4/zephyr-orpo-141b-A35b-v0.1",
+    # "openchat-3.5": "openchat/openchat-3.5-0106",
+    # "command-r-plus": "CohereForAI/c4ai-command-r-plus",
+    # "llama3-70b": "meta-llama/Meta-Llama-3-70B-Instruct",
+    # "zephyr-141b": "HuggingFaceH4/zephyr-orpo-141b-A35b-v0.1",
     "default": "NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO",
 }
 
@@ -15,24 +16,31 @@ AVAILABLE_MODELS = list(MODEL_MAP.keys())
 PRO_MODELS = ["command-r-plus", "llama3-70b", "zephyr-141b"]
 
 STOP_SEQUENCES_MAP = {
+    # https://huggingface.co/mistralai/Mixtral-8x7B-Instruct-v0.1/blob/main/tokenizer_config.json#L33
     "mixtral-8x7b": "</s>",
+    # https://huggingface.co/NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO/blob/main/tokenizer_config.json#L50
     "nous-mixtral-8x7b": "<|im_end|>",
+    # https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2/blob/main/tokenizer_config.json#L33
     "mistral-7b": "</s>",
-    # "openchat-3.5": "<|end_of_turn|>",
+    # https://huggingface.co/01-ai/Yi-1.5-34B-Chat/blob/main/tokenizer_config.json#L42
+    "yi-1.5-34b": "<|im_end|>",
+    # https://huggingface.co/google/gemma-1.1-7b-it/blob/main/tokenizer_config.json#L1509
     "gemma-7b": "<eos>",
-    "command-r-plus": "<|END_OF_TURN_TOKEN|>",
+    # "openchat-3.5": "<|end_of_turn|>",
+    # "command-r-plus": "<|END_OF_TURN_TOKEN|>",
 }
 
 TOKEN_LIMIT_MAP = {
     "mixtral-8x7b": 32768,
     "nous-mixtral-8x7b": 32768,
     "mistral-7b": 32768,
-    # "openchat-3.5": 8192,
+    "yi-1.5-34b": 4096,
     "gemma-7b": 8192,
-    "command-r-plus": 32768,
-    "llama3-70b": 8192,
-    "zephyr-141b": 2048,
-    "gpt-3.5-turbo": 8192,
+    # "openchat-3.5": 8192,
+    # "command-r-plus": 32768,
+    # "llama3-70b": 8192,
+    # "zephyr-141b": 2048,
+    # "gpt-3.5-turbo": 8192,
 }
 
 TOKEN_RESERVED = 20
@@ -61,46 +69,53 @@ AVAILABLE_MODELS_DICTS = [
         "created": 1700000000,
         "owned_by": "mistralai",
     },
-    # {
-    #     "id": "openchat-3.5",
-    #     "description": "[openchat/openchat-3.5-0106]: https://huggingface.co/openchat/openchat-3.5-0106",
-    #     "object": "model",
-    #     "created": 1700000000,
-    #     "owned_by": "openchat",
-    # },
-    {
-        "id": "gemma-7b",
-        "description": "[google/gemma-1.1-7b-it]: https://huggingface.co/google/gemma-1.1-7b-it",
-        "object": "model",
-        "created": 1700000000,
-        "owned_by": "Google",
-    },
-    {
-        "id": "command-r-plus",
-        "description": "[CohereForAI/c4ai-command-r-plus]: https://huggingface.co/CohereForAI/c4ai-command-r-plus",
-        "object": "model",
-        "created": 1700000000,
-        "owned_by": "CohereForAI",
-    },
-    {
-        "id": "llama3-70b",
-        "description": "[meta-llama/Meta-Llama-3-70B]: https://huggingface.co/meta-llama/Meta-Llama-3-70B",
-        "object": "model",
-        "created": 1700000000,
-        "owned_by": "Meta",
-    },
     {
-        "id": "zephyr-141b",
-        "description": "[HuggingFaceH4/zephyr-orpo-141b-A35b-v0.1]: https://huggingface.co/HuggingFaceH4/zephyr-orpo-141b-A35b-v0.1",
+        "id": "yi-1.5-34b",
+        "description": "[01-ai/Yi-1.5-34B-Chat]: https://huggingface.co/01-ai/Yi-1.5-34B-Chat",
         "object": "model",
         "created": 1700000000,
-        "owned_by": "Huggingface",
+        "owned_by": "01-ai",
     },
     {
-        "id": "gpt-3.5-turbo",
-        "description": "[openai/gpt-3.5-turbo]: https://platform.openai.com/docs/models/gpt-3-5-turbo",
+        "id": "gemma-7b",
+        "description": "[google/gemma-1.1-7b-it]: https://huggingface.co/google/gemma-1.1-7b-it",
         "object": "model",
         "created": 1700000000,
-        "owned_by": "OpenAI",
+        "owned_by": "Google",
     },
+    # {
+    #     "id": "openchat-3.5",
+    #     "description": "[openchat/openchat-3.5-0106]: https://huggingface.co/openchat/openchat-3.5-0106",
+    #     "object": "model",
+    #     "created": 1700000000,
+    #     "owned_by": "openchat",
+    # },
+    # {
+    #     "id": "command-r-plus",
+    #     "description": "[CohereForAI/c4ai-command-r-plus]: https://huggingface.co/CohereForAI/c4ai-command-r-plus",
+    #     "object": "model",
+    #     "created": 1700000000,
+    #     "owned_by": "CohereForAI",
+    # },
+    # {
+    #     "id": "llama3-70b",
+    #     "description": "[meta-llama/Meta-Llama-3-70B]: https://huggingface.co/meta-llama/Meta-Llama-3-70B",
+    #     "object": "model",
+    #     "created": 1700000000,
+    #     "owned_by": "Meta",
+    # },
+    # {
+    #     "id": "zephyr-141b",
+    #     "description": "[HuggingFaceH4/zephyr-orpo-141b-A35b-v0.1]: https://huggingface.co/HuggingFaceH4/zephyr-orpo-141b-A35b-v0.1",
+    #     "object": "model",
+    #     "created": 1700000000,
+    #     "owned_by": "Huggingface",
+    # },
+    # {
+    #     "id": "gpt-3.5-turbo",
+    #     "description": "[openai/gpt-3.5-turbo]: https://platform.openai.com/docs/models/gpt-3-5-turbo",
+    #     "object": "model",
+    #     "created": 1700000000,
+    #     "owned_by": "OpenAI",
+    # },
 ]

messagers/message_composer.py

@@ -150,8 +150,8 @@ class MessageComposer:
             self.merged_str = "<bos>" + "\n".join(self.merged_str_list)
         # https://huggingface.co/NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO#prompt-format
         # https://huggingface.co/openchat/openchat-3.5-0106
-        # elif self.model in ["openchat-3.5", "nous-mixtral-8x7b"]:
-        elif self.model in ["openchat-3.5", "command-r-plus", "gemma-7b"]:
+        # https://huggingface.co/01-ai/Yi-1.5-34B-Chat
+        elif self.model in ["openchat-3.5", "command-r-plus", "gemma-7b", "yi-1.5-34b"]:
             tokenizer = AutoTokenizer.from_pretrained(self.model_fullname)
             self.merged_str = tokenizer.apply_chat_template(
                 messages, tokenize=False, add_generation_prompt=True