diff --git a/server/app/type/model_providers.py b/server/app/type/model_providers.py deleted file mode 100644 index c9fab31f..00000000 --- a/server/app/type/model_providers.py +++ /dev/null @@ -1,64 +0,0 @@ -from enum import Enum -from typing import List - - -class ModelProviders(Enum): - OPENAI = "openai" - AWS_BEDROCK = "aws-bedrock" - AZURE = "azure" - ANTHROPIC = "anthropic" - GROQ = "groq" - OPENROUTER = "openrouter" - OLLAMA = "ollama" - LITELLM = "litellm" - LMSTUDIO = "lmstudio" - ZHIPU = "zhipuai" - GEMINI = "gemini" - VLLM = "vllm" - MISTRAL = "mistral" - REKA = "reka" - TOGETHER = "together" - STUB = "stub" - OPENAI_COMPATIBLE_MODEL = "openai-compatible-model" - SAMBA = "samba-nova" - COHERE = "cohere" - YI = "lingyiwanwu" - QWEN = "tongyi-qianwen" - NVIDIA = "nvidia" - DEEPSEEK = "deepseek" - PPIO = "ppio" - SGLANG = "sglang" - INTERNLM = "internlm" - MOONSHOT = "moonshot" - MODELSCOPE = "modelscope" - SILICONFLOW = "siliconflow" - AIML = "aiml" - VOLCANO = "volcano" - NETMIND = "netmind" - NOVITA = "novita" - WATSONX = "watsonx" - MINIMAX = "minimax" - - @classmethod - def get_all_values(cls) -> List[str]: - return [platform.value for platform in cls] - - @classmethod - def get_all_names(cls) -> List[str]: - return [platform.name for platform in cls] - - @classmethod - def get_all_items(cls) -> List[dict]: - return [{"name": platform.name, "value": platform.value} for platform in cls] - - @classmethod - def is_valid_platform(cls, platform_name: str) -> bool: - try: - cls(platform_name) - return True - except ValueError: - return False - - @classmethod - def get_platform_by_name(cls, platform_name: str) -> "ModelPlatformType": - return cls(platform_name) diff --git a/server/app/type/model_type.py b/server/app/type/model_type.py deleted file mode 100644 index 8fc1abd7..00000000 --- a/server/app/type/model_type.py +++ /dev/null @@ -1,344 +0,0 @@ -from enum import Enum - - -class ModelType(Enum): - GPT_3_5_TURBO = "gpt-3.5-turbo" - GPT_4 = "gpt-4" - GPT_4_TURBO = "gpt-4-turbo" - GPT_4O = "gpt-4o" - GPT_4O_MINI = "gpt-4o-mini" - GPT_4_5_PREVIEW = "gpt-4.5-preview" - O1 = "o1" - O1_PREVIEW = "o1-preview" - O1_MINI = "o1-mini" - O3_MINI = "o3-mini" - GPT_4_1 = "gpt-4.1-2025-04-14" - GPT_4_1_MINI = "gpt-4.1-mini-2025-04-14" - GPT_4_1_NANO = "gpt-4.1-nano-2025-04-14" - O4_MINI = "o4-mini" - O3 = "o3" - O3_PRO = "o3-pro" - - AWS_CLAUDE_3_7_SONNET = "anthropic.claude-3-7-sonnet-20250219-v1:0" - AWS_CLAUDE_3_5_SONNET = "anthropic.claude-3-5-sonnet-20241022-v2:0" - AWS_CLAUDE_3_HAIKU = "anthropic.claude-3-haiku-20240307-v1:0" - AWS_CLAUDE_3_SONNET = "anthropic.claude-3-sonnet-20240229-v1:0" - AWS_DEEPSEEK_R1 = "us.deepseek.r1-v1:0" - AWS_LLAMA_3_3_70B_INSTRUCT = "us.meta.llama3-3-70b-instruct-v1:0" - AWS_LLAMA_3_2_90B_INSTRUCT = "us.meta.llama3-2-90b-instruct-v1:0" - AWS_LLAMA_3_2_11B_INSTRUCT = "us.meta.llama3-2-11b-instruct-v1:0" - AWS_CLAUDE_SONNET_4 = "anthropic.claude-sonnet-4-20250514-v1:0" - AWS_CLAUDE_OPUS_4 = "anthropic.claude-opus-4-20250514-v1:0" - - GLM_4 = "glm-4" - GLM_4V = "glm-4v" - GLM_4V_FLASH = "glm-4v-flash" - GLM_4V_PLUS_0111 = "glm-4v-plus-0111" - GLM_4_PLUS = "glm-4-plus" - GLM_4_AIR = "glm-4-air" - GLM_4_AIR_0111 = "glm-4-air-0111" - GLM_4_AIRX = "glm-4-airx" - GLM_4_LONG = "glm-4-long" - GLM_4_FLASHX = "glm-4-flashx" - GLM_4_FLASH = "glm-4-flash" - GLM_ZERO_PREVIEW = "glm-zero-preview" - GLM_3_TURBO = "glm-3-turbo" - - # Groq platform models - GROQ_LLAMA_3_1_8B = "llama-3.1-8b-instant" - GROQ_LLAMA_3_3_70B = "llama-3.3-70b-versatile" - GROQ_LLAMA_3_3_70B_PREVIEW = "llama-3.3-70b-specdec" - GROQ_LLAMA_3_8B = "llama3-8b-8192" - GROQ_LLAMA_3_70B = "llama3-70b-8192" - GROQ_MIXTRAL_8_7B = "mixtral-8x7b-32768" - GROQ_GEMMA_2_9B_IT = "gemma2-9b-it" - - # OpenRouter models - OPENROUTER_LLAMA_3_1_405B = "meta-llama/llama-3.1-405b-instruct" - OPENROUTER_LLAMA_3_1_70B = "meta-llama/llama-3.1-70b-instruct" - OPENROUTER_LLAMA_4_MAVERICK = "meta-llama/llama-4-maverick" - OPENROUTER_LLAMA_4_MAVERICK_FREE = "meta-llama/llama-4-maverick:free" - OPENROUTER_LLAMA_4_SCOUT = "meta-llama/llama-4-scout" - OPENROUTER_LLAMA_4_SCOUT_FREE = "meta-llama/llama-4-scout:free" - OPENROUTER_OLYMPICODER_7B = "open-r1/olympiccoder-7b:free" - - # LMStudio models - LMSTUDIO_GEMMA_3_1B = "gemma-3-1b" - LMSTUDIO_GEMMA_3_4B = "gemma-3-4b" - LMSTUDIO_GEMMA_3_12B = "gemma-3-12b" - LMSTUDIO_GEMMA_3_27B = "gemma-3-27b" - - # TogetherAI platform models support tool calling - TOGETHER_LLAMA_3_1_8B = "meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo" - TOGETHER_LLAMA_3_1_70B = "meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo" - TOGETHER_LLAMA_3_1_405B = "meta-llama/Meta-Llama-3.1-405B-Instruct-Turbo" - TOGETHER_LLAMA_3_3_70B = "meta-llama/Llama-3.3-70B-Instruct-Turbo" - TOGETHER_MIXTRAL_8_7B = "mistralai/Mixtral-8x7B-Instruct-v0.1" - TOGETHER_MISTRAL_7B = "mistralai/Mistral-7B-Instruct-v0.1" - TOGETHER_LLAMA_4_MAVERICK = "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8" - TOGETHER_LLAMA_4_SCOUT = "meta-llama/Llama-4-Scout-17B-16E-Instruct" - - # PPIO platform models support tool calling - PPIO_DEEPSEEK_PROVER_V2_671B = "deepseek/deepseek-prover-v2-671b" - PPIO_DEEPSEEK_R1_TURBO = "deepseek/deepseek-r1-turbo" - PPIO_DEEPSEEK_V3_TURBO = "deepseek/deepseek-v3-turbo" - PPIO_DEEPSEEK_R1_COMMUNITY = "deepseek/deepseek-r1/community" - PPIO_DEEPSEEK_V3_COMMUNITY = "deepseek/deepseek-v3/community" - PPIO_DEEPSEEK_R1 = "deepseek/deepseek-r1" - PPIO_DEEPSEEK_V3 = "deepseek/deepseek-v3" - PPIO_QWEN_2_5_72B = "qwen/qwen-2.5-72b-instruct" - PPIO_BAICHUAN_2_13B_CHAT = "baichuan/baichuan2-13b-chat" - PPIO_LLAMA_3_3_70B = "meta-llama/llama-3.3-70b-instruct" - PPIO_LLAMA_3_1_70B = "meta-llama/llama-3.1-70b-instruct" - PPIO_YI_1_5_34B_CHAT = "01-ai/yi-1.5-34b-chat" - - # SambaNova Cloud platform models support tool calling - SAMBA_LLAMA_3_1_8B = "Meta-Llama-3.1-8B-Instruct" - SAMBA_LLAMA_3_1_70B = "Meta-Llama-3.1-70B-Instruct" - SAMBA_LLAMA_3_1_405B = "Meta-Llama-3.1-405B-Instruct" - - # SGLang models support tool calling - SGLANG_LLAMA_3_1_8B = "meta-llama/Meta-Llama-3.1-8B-Instruct" - SGLANG_LLAMA_3_1_70B = "meta-llama/Meta-Llama-3.1-70B-Instruct" - SGLANG_LLAMA_3_1_405B = "meta-llama/Meta-Llama-3.1-405B-Instruct" - SGLANG_LLAMA_3_2_1B = "meta-llama/Llama-3.2-1B-Instruct" - SGLANG_MIXTRAL_NEMO = "mistralai/Mistral-Nemo-Instruct-2407" - SGLANG_MISTRAL_7B = "mistralai/Mistral-7B-Instruct-v0.3" - SGLANG_QWEN_2_5_7B = "Qwen/Qwen2.5-7B-Instruct" - SGLANG_QWEN_2_5_32B = "Qwen/Qwen2.5-32B-Instruct" - SGLANG_QWEN_2_5_72B = "Qwen/Qwen2.5-72B-Instruct" - - STUB = "stub" - - # Legacy anthropic models - # NOTE: anthropic legacy models only Claude 2.1 has system prompt support - CLAUDE_2_1 = "claude-2.1" - CLAUDE_2_0 = "claude-2.0" - CLAUDE_INSTANT_1_2 = "claude-instant-1.2" - - # Claude models - CLAUDE_3_OPUS = "claude-3-opus-latest" - CLAUDE_3_SONNET = "claude-3-sonnet-20240229" - CLAUDE_3_HAIKU = "claude-3-haiku-20240307" - CLAUDE_3_5_SONNET = "claude-3-5-sonnet-latest" - CLAUDE_3_5_HAIKU = "claude-3-5-haiku-latest" - CLAUDE_3_7_SONNET = "claude-3-7-sonnet-latest" - CLAUDE_SONNET_4 = "claude-sonnet-4-20250514" - CLAUDE_OPUS_4 = "claude-opus-4-20250514" - - # Netmind models - NETMIND_LLAMA_4_MAVERICK_17B_128E_INSTRUCT = "meta-llama/Llama-4-Maverick-17B-128E-Instruct" - NETMIND_LLAMA_4_SCOUT_17B_16E_INSTRUCT = "meta-llama/Llama-4-Scout-17B-16E-Instruct" - NETMIND_DEEPSEEK_R1 = "deepseek-ai/DeepSeek-R1" - NETMIND_DEEPSEEK_V3 = "deepseek-ai/DeepSeek-V3-0324" - NETMIND_DOUBAO_1_5_PRO = "doubao/Doubao-1.5-pro" - NETMIND_QWQ_32B = "Qwen/QwQ-32B" - - # Nvidia models - NVIDIA_NEMOTRON_340B_INSTRUCT = "nvidia/nemotron-4-340b-instruct" - NVIDIA_NEMOTRON_340B_REWARD = "nvidia/nemotron-4-340b-reward" - NVIDIA_YI_LARGE = "01-ai/yi-large" - NVIDIA_MISTRAL_LARGE = "mistralai/mistral-large" - NVIDIA_MIXTRAL_8X7B = "mistralai/mixtral-8x7b-instruct" - NVIDIA_LLAMA3_70B = "meta/llama3-70b" - NVIDIA_LLAMA3_1_8B_INSTRUCT = "meta/llama-3.1-8b-instruct" - NVIDIA_LLAMA3_1_70B_INSTRUCT = "meta/llama-3.1-70b-instruct" - NVIDIA_LLAMA3_1_405B_INSTRUCT = "meta/llama-3.1-405b-instruct" - NVIDIA_LLAMA3_2_1B_INSTRUCT = "meta/llama-3.2-1b-instruct" - NVIDIA_LLAMA3_2_3B_INSTRUCT = "meta/llama-3.2-3b-instruct" - NVIDIA_LLAMA3_3_70B_INSTRUCT = "meta/llama-3.3-70b-instruct" - - # Gemini models - GEMINI_3_PRO ="gemini-3-pro-preview" - GEMINI_2_5_FLASH_PREVIEW = "gemini-2.5-flash-preview-04-17" - GEMINI_2_5_PRO_PREVIEW = "gemini-2.5-pro-preview-06-05" - GEMINI_2_0_FLASH = "gemini-2.0-flash" - GEMINI_2_0_FLASH_EXP = "gemini-2.0-flash-exp" - GEMINI_2_0_FLASH_THINKING = "gemini-2.0-flash-thinking-exp" - GEMINI_2_0_PRO_EXP = "gemini-2.0-pro-exp-02-05" - GEMINI_2_0_FLASH_LITE = "gemini-2.0-flash-lite" - GEMINI_2_0_FLASH_LITE_PREVIEW = "gemini-2.0-flash-lite-preview-02-05" - GEMINI_1_5_FLASH = "gemini-1.5-flash" - GEMINI_1_5_PRO = "gemini-1.5-pro" - - # Mistral AI models - MISTRAL_3B = "ministral-3b-latest" - MISTRAL_7B = "open-mistral-7b" - MISTRAL_8B = "ministral-8b-latest" - MISTRAL_CODESTRAL = "codestral-latest" - MISTRAL_CODESTRAL_MAMBA = "open-codestral-mamba" - MISTRAL_LARGE = "mistral-large-latest" - MISTRAL_MIXTRAL_8x7B = "open-mixtral-8x7b" - MISTRAL_MIXTRAL_8x22B = "open-mixtral-8x22b" - MISTRAL_NEMO = "open-mistral-nemo" - MISTRAL_PIXTRAL_12B = "pixtral-12b-2409" - MISTRAL_MEDIUM_3 = "mistral-medium-latest" - MAGISTRAL_MEDIUM = "magistral-medium-2506" - - # Reka models - REKA_CORE = "reka-core" - REKA_FLASH = "reka-flash" - REKA_EDGE = "reka-edge" - - # Cohere models - COHERE_COMMAND_R_PLUS = "command-r-plus" - COHERE_COMMAND_R = "command-r" - COHERE_COMMAND_LIGHT = "command-light" - COHERE_COMMAND = "command" - COHERE_COMMAND_NIGHTLY = "command-nightly" - - # Qwen models (Aliyun) - QWEN_MAX = "qwen-max" - QWEN_PLUS = "qwen-plus" - QWEN_TURBO = "qwen-turbo" - QWEN_PLUS_LATEST = "qwen-plus-latest" - QWEN_PLUS_2025_04_28 = "qwen-plus-2025-04-28" - QWEN_TURBO_LATEST = "qwen-turbo-latest" - QWEN_TURBO_2025_04_28 = "qwen-turbo-2025-04-28" - QWEN_LONG = "qwen-long" - QWEN_VL_MAX = "qwen-vl-max" - QWEN_VL_PLUS = "qwen-vl-plus" - QWEN_MATH_PLUS = "qwen-math-plus" - QWEN_MATH_TURBO = "qwen-math-turbo" - QWEN_CODER_TURBO = "qwen-coder-turbo" - QWEN_2_5_CODER_32B = "qwen2.5-coder-32b-instruct" - QWEN_2_5_VL_72B = "qwen2.5-vl-72b-instruct" - QWEN_2_5_72B = "qwen2.5-72b-instruct" - QWEN_2_5_32B = "qwen2.5-32b-instruct" - QWEN_2_5_14B = "qwen2.5-14b-instruct" - QWEN_QWQ_32B = "qwq-32b-preview" - QWEN_QVQ_72B = "qvq-72b-preview" - QWEN_QWQ_PLUS = "qwq-plus" - - # Yi models (01-ai) - YI_LIGHTNING = "yi-lightning" - YI_LARGE = "yi-large" - YI_MEDIUM = "yi-medium" - YI_LARGE_TURBO = "yi-large-turbo" - YI_VISION = "yi-vision" - YI_MEDIUM_200K = "yi-medium-200k" - YI_SPARK = "yi-spark" - YI_LARGE_RAG = "yi-large-rag" - YI_LARGE_FC = "yi-large-fc" - - # DeepSeek models - DEEPSEEK_CHAT = "deepseek-chat" - DEEPSEEK_REASONER = "deepseek-reasoner" - # InternLM models - INTERNLM3_LATEST = "internlm3-latest" - INTERNLM3_8B_INSTRUCT = "internlm3-8b-instruct" - INTERNLM2_5_LATEST = "internlm2.5-latest" - INTERNLM2_PRO_CHAT = "internlm2-pro-chat" - - # Moonshot models - MOONSHOT_V1_8K = "moonshot-v1-8k" - MOONSHOT_V1_32K = "moonshot-v1-32k" - MOONSHOT_V1_128K = "moonshot-v1-128k" - - # SiliconFlow models support tool calling - SILICONFLOW_DEEPSEEK_V2_5 = "deepseek-ai/DeepSeek-V2.5" - SILICONFLOW_DEEPSEEK_V3 = "deepseek-ai/DeepSeek-V3" - SILICONFLOW_INTERN_LM2_5_20B_CHAT = "internlm/internlm2_5-20b-chat" - SILICONFLOW_INTERN_LM2_5_7B_CHAT = "internlm/internlm2_5-7b-chat" - SILICONFLOW_PRO_INTERN_LM2_5_7B_CHAT = "Pro/internlm/internlm2_5-7b-chat" - SILICONFLOW_QWEN2_5_72B_INSTRUCT = "Qwen/Qwen2.5-72B-Instruct" - SILICONFLOW_QWEN2_5_32B_INSTRUCT = "Qwen/Qwen2.5-32B-Instruct" - SILICONFLOW_QWEN2_5_14B_INSTRUCT = "Qwen/Qwen2.5-14B-Instruct" - SILICONFLOW_QWEN2_5_7B_INSTRUCT = "Qwen/Qwen2.5-7B-Instruct" - SILICONFLOW_PRO_QWEN2_5_7B_INSTRUCT = "Pro/Qwen/Qwen2.5-7B-Instruct" - SILICONFLOW_THUDM_GLM_4_9B_CHAT = "THUDM/glm-4-9b-chat" - SILICONFLOW_PRO_THUDM_GLM_4_9B_CHAT = "Pro/THUDM/glm-4-9b-chat" - - # AIML models support tool calling - AIML_MIXTRAL_8X7B = "mistralai/Mixtral-8x7B-Instruct-v0.1" - AIML_MISTRAL_7B_INSTRUCT = "mistralai/Mistral-7B-Instruct-v0.1" - - # Novita platform models support tool calling - NOVITA_LLAMA_4_MAVERICK_17B = "meta-llama/llama-4-maverick-17b-128e-instruct-fp8" - NOVITA_LLAMA_4_SCOUT_17B = "meta-llama/llama-4-scout-17b-16e-instruct" - NOVITA_DEEPSEEK_V3_0324 = "deepseek/deepseek-v3-0324" - NOVITA_QWEN_2_5_V1_72B = "qwen/qwen2.5-vl-72b-instruct" - NOVITA_DEEPSEEK_V3_TURBO = "deepseek/deepseek-v3-turbo" - NOVITA_DEEPSEEK_R1_TURBO = "deepseek/deepseek-r1-turbo" - NOVITA_GEMMA_3_27B_IT = "google/gemma-3-27b-it" - NOVITA_QWEN_32B = "qwen/qwq-32b" - NOVITA_L3_8B_STHENO_V3_2 = "Sao10K/L3-8B-Stheno-v3.2" - NOVITA_MYTHOMAX_L2_13B = "gryphe/mythomax-l2-13b" - NOVITA_DEEPSEEK_R1_DISTILL_LLAMA_8B = "deepseek/deepseek-r1-distill-llama-8b" - NOVITA_DEEPSEEK_V3 = "deepseek/deepseek_v3" - NOVITA_LLAMA_3_1_8B = "meta-llama/llama-3.1-8b-instruct" - NOVITA_DEEPSEEK_R1_DISTILL_QWEN_14B = "deepseek/deepseek-r1-distill-qwen-14b" - NOVITA_LLAMA_3_3_70B = "meta-llama/llama-3.3-70b-instruct" - NOVITA_QWEN_2_5_72B = "qwen/qwen-2.5-72b-instruct" - NOVITA_MISTRAL_NEMO = "mistralai/mistral-nemo" - NOVITA_DEEPSEEK_R1_DISTILL_QWEN_32B = "deepseek/deepseek-r1-distill-qwen-32b" - NOVITA_LLAMA_3_8B = "meta-llama/llama-3-8b-instruct" - NOVITA_WIZARDLM_2_8X22B = "microsoft/wizardlm-2-8x22b" - NOVITA_DEEPSEEK_R1_DISTILL_LLAMA_70B = "deepseek/deepseek-r1-distill-llama-70b" - NOVITA_LLAMA_3_1_70B = "meta-llama/llama-3.1-70b-instruct" - NOVITA_GEMMA_2_9B_IT = "google/gemma-2-9b-it" - NOVITA_MISTRAL_7B = "mistralai/mistral-7b-instruct" - NOVITA_LLAMA_3_70B = "meta-llama/llama-3-70b-instruct" - NOVITA_DEEPSEEK_R1 = "deepseek/deepseek-r1" - NOVITA_HERMES_2_PRO_LLAMA_3_8B = "nousresearch/hermes-2-pro-llama-3-8b" - NOVITA_L3_70B_EURYALE_V2_1 = "sao10k/l3-70b-euryale-v2.1" - NOVITA_DOLPHIN_MIXTRAL_8X22B = "cognitivecomputations/dolphin-mixtral-8x22b" - NOVITA_AIROBOROS_L2_70B = "jondurbin/airoboros-l2-70b" - NOVITA_MIDNIGHT_ROSE_70B = "sophosympatheia/midnight-rose-70b" - NOVITA_L3_8B_LUNARIS = "sao10k/l3-8b-lunaris" - NOVITA_GLM_4_9B_0414 = "thudm/glm-4-9b-0414" - NOVITA_GLM_Z1_9B_0414 = "thudm/glm-z1-9b-0414" - NOVITA_GLM_Z1_32B_0414 = "thudm/glm-z1-32b-0414" - NOVITA_GLM_4_32B_0414 = "thudm/glm-4-32b-0414" - NOVITA_GLM_Z1_RUMINATION_32B_0414 = "thudm/glm-z1-rumination-32b-0414" - NOVITA_QWEN_2_5_7B = "qwen/qwen2.5-7b-instruct" - NOVITA_LLAMA_3_2_1B = "meta-llama/llama-3.2-1b-instruct" - NOVITA_LLAMA_3_2_11B_VISION = "meta-llama/llama-3.2-11b-vision-instruct" - NOVITA_LLAMA_3_2_3B = "meta-llama/llama-3.2-3b-instruct" - NOVITA_LLAMA_3_1_8B_BF16 = "meta-llama/llama-3.1-8b-instruct-bf16" - NOVITA_L31_70B_EURYALE_V2_2 = "sao10k/l31-70b-euryale-v2.2" - - # ModelScope models support tool calling - MODELSCOPE_QWEN_2_5_7B_INSTRUCT = "Qwen/Qwen2.5-7B-Instruct" - MODELSCOPE_QWEN_2_5_14B_INSTRUCT = "Qwen/Qwen2.5-14B-Instruct" - MODELSCOPE_QWEN_2_5_32B_INSTRUCT = "Qwen/Qwen2.5-32B-Instruct" - MODELSCOPE_QWEN_2_5_72B_INSTRUCT = "Qwen/Qwen2.5-72B-Instruct" - MODELSCOPE_QWEN_2_5_CODER_7B_INSTRUCT = "Qwen/Qwen2.5-Coder-7B-Instruct" - MODELSCOPE_QWEN_2_5_CODER_14B_INSTRUCT = "Qwen/Qwen2.5-Coder-14B-Instruct" - MODELSCOPE_QWEN_2_5_CODER_32B_INSTRUCT = "Qwen/Qwen2.5-Coder-32B-Instruct" - MODELSCOPE_QWEN_3_235B_A22B = "Qwen/Qwen3-235B-A22B" - MODELSCOPE_QWEN_3_32B = "Qwen/Qwen3-32B" - MODELSCOPE_QWQ_32B = "Qwen/QwQ-32B" - MODELSCOPE_QWQ_32B_PREVIEW = "Qwen/QwQ-32B-Preview" - MODELSCOPE_LLAMA_3_1_8B_INSTRUCT = "LLM-Research/Meta-Llama-3.1-8B-Instruct" - MODELSCOPE_LLAMA_3_1_70B_INSTRUCT = "LLM-Research/Meta-Llama-3.1-70B-Instruct" - MODELSCOPE_LLAMA_3_1_405B_INSTRUCT = "LLM-Research/Meta-Llama-3.1-405B-Instruct" - MODELSCOPE_LLAMA_3_3_70B_INSTRUCT = "LLM-Research/Llama-3.3-70B-Instruct" - MODELSCOPE_MINISTRAL_8B_INSTRUCT = "mistralai/Ministral-8B-Instruct-2410" - MODELSCOPE_DEEPSEEK_V3_0324 = "deepseek-ai/DeepSeek-V3-0324" - - # WatsonX models - WATSONX_GRANITE_3_8B_INSTRUCT = "ibm/granite-3-8b-instruct" - WATSONX_LLAMA_3_3_70B_INSTRUCT = "meta-llama/llama-3-3-70b-instruct" - WATSONX_LLAMA_3_2_1B_INSTRUCT = "meta-llama/llama-3-2-1b-instruct" - WATSONX_LLAMA_3_2_3B_INSTRUCT = "meta-llama/llama-3-2-3b-instruct" - WATSONX_LLAMA_3_2_11B_VISION_INSTRUCT = "meta-llama/llama-3-2-11b-vision-instruct" - WATSONX_LLAMA_3_2_90B_VISION_INSTRUCT = "meta-llama/llama-3-2-90b-vision-instruct" - WATSONX_LLAMA_GUARD_3_11B_VISION_INSTRUCT = "meta-llama/llama-guard-3-11b-vision-instruct" - WATSONX_MISTRAL_LARGE = "mistralai/mistral-large" - - # Crynux models - CRYNUX_DEEPSEEK_R1_DISTILL_QWEN_1_5B = "deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B" - CRYNUX_DEEPSEEK_R1_DISTILL_QWEN_7B = "deepseek-ai/DeepSeek-R1-Distill-Qwen-7B" - CRYNUX_DEEPSEEK_R1_DISTILL_LLAMA_8B = "deepseek-ai/DeepSeek-R1-Distill-Llama-8B" - - CRYNUX_QWEN_3_4_B = "Qwen/Qwen3-4B" - CRYNUX_QWEN_3_8_B = "Qwen/Qwen3-8B" - CRYNUX_QWEN_2_5_7B = "Qwen/Qwen2.5-7B" - CRYNUX_QWEN_2_5_7B_INSTRUCT = "Qwen/Qwen2.5-7B-Instruct" - - CRYNUX_NOUS_HERMES_3_LLAMA_3_1_8B = "NousResearch/Hermes-3-Llama-3.1-8B" - CRYNUX_NOUS_HERMES_3_LLAMA_3_2_3B = "NousResearch/Hermes-3-Llama-3.2-3B" - - def __str__(self): - return self.value