135 lines
4.3 KiB
TOML
135 lines
4.3 KiB
TOML
[inner]
|
||
version = "0.1.1"
|
||
|
||
# 配置文件版本号迭代规则同bot_config.toml
|
||
|
||
[request_conf] # 请求配置（此配置项数值均为默认值，如想修改，请取消对应条目的注释）
|
||
#max_retry = 2 # 最大重试次数（单个模型API调用失败，最多重试的次数）
|
||
#timeout = 10 # API调用的超时时长（超过这个时长，本次请求将被视为“请求超时”，单位：秒）
|
||
#retry_interval = 10 # 重试间隔（如果API调用失败，重试的间隔时间，单位：秒）
|
||
#default_temperature = 0.7 # 默认的温度（如果bot_config.toml中没有设置temperature参数，默认使用这个值）
|
||
#default_max_tokens = 1024 # 默认的最大输出token数（如果bot_config.toml中没有设置max_tokens参数，默认使用这个值）
|
||
|
||
|
||
[[api_providers]] # API服务提供商（可以配置多个）
|
||
name = "DeepSeek" # API服务商名称（可随意命名，在models的api-provider中需使用这个命名）
|
||
base_url = "https://api.deepseek.cn" # API服务商的BaseURL
|
||
key = "******" # API Key （可选，默认为None）
|
||
client_type = "openai" # 请求客户端（可选，默认值为"openai"，使用gimini等Google系模型时请配置为"google"）
|
||
|
||
#[[api_providers]] # 特殊：Google的Gimini使用特殊API，与OpenAI格式不兼容，需要配置client为"google"
|
||
#name = "Google"
|
||
#base_url = "https://api.google.com"
|
||
#key = "******"
|
||
#client_type = "google"
|
||
#
|
||
#[[api_providers]]
|
||
#name = "SiliconFlow"
|
||
#base_url = "https://api.siliconflow.cn"
|
||
#key = "******"
|
||
#
|
||
#[[api_providers]]
|
||
#name = "LocalHost"
|
||
#base_url = "https://localhost:8888"
|
||
#key = "lm-studio"
|
||
|
||
|
||
[[models]] # 模型（可以配置多个）
|
||
# 模型标识符（API服务商提供的模型标识符）
|
||
model_identifier = "deepseek-chat"
|
||
# 模型名称（可随意命名，在bot_config.toml中需使用这个命名）
|
||
#（可选，若无该字段，则将自动使用model_identifier填充）
|
||
name = "deepseek-v3"
|
||
# API服务商名称（对应在api_providers中配置的服务商名称）
|
||
api_provider = "DeepSeek"
|
||
# 输入价格（用于API调用统计，单位：元/兆token）（可选，若无该字段，默认值为0）
|
||
price_in = 2.0
|
||
# 输出价格（用于API调用统计，单位：元/兆token）（可选，若无该字段，默认值为0）
|
||
price_out = 8.0
|
||
# 强制流式输出模式（若模型不支持非流式输出，请取消该注释，启用强制流式输出）
|
||
#（可选，若无该字段，默认值为false）
|
||
#force_stream_mode = true
|
||
|
||
[[models]]
|
||
model_identifier = "deepseek-reasoner"
|
||
name = "deepseek-r1"
|
||
api_provider = "DeepSeek"
|
||
model_flags = [ "text", "tool_calling", "reasoning",]
|
||
price_in = 4.0
|
||
price_out = 16.0
|
||
|
||
[[models]]
|
||
model_identifier = "Pro/deepseek-ai/DeepSeek-V3"
|
||
name = "siliconflow-deepseek-v3"
|
||
api_provider = "SiliconFlow"
|
||
price_in = 2.0
|
||
price_out = 8.0
|
||
|
||
[[models]]
|
||
model_identifier = "Pro/deepseek-ai/DeepSeek-R1"
|
||
name = "siliconflow-deepseek-r1"
|
||
api_provider = "SiliconFlow"
|
||
price_in = 4.0
|
||
price_out = 16.0
|
||
|
||
[[models]]
|
||
model_identifier = "Pro/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B"
|
||
name = "deepseek-r1-distill-qwen-32b"
|
||
api_provider = "SiliconFlow"
|
||
price_in = 4.0
|
||
price_out = 16.0
|
||
|
||
[[models]]
|
||
model_identifier = "Qwen/Qwen3-8B"
|
||
name = "qwen3-8b"
|
||
api_provider = "SiliconFlow"
|
||
price_in = 0
|
||
price_out = 0
|
||
|
||
[[models]]
|
||
model_identifier = "Qwen/Qwen3-14B"
|
||
name = "qwen3-14b"
|
||
api_provider = "SiliconFlow"
|
||
price_in = 0.5
|
||
price_out = 2.0
|
||
|
||
[[models]]
|
||
model_identifier = "Qwen/Qwen3-30B-A3B"
|
||
name = "qwen3-30b"
|
||
api_provider = "SiliconFlow"
|
||
price_in = 0.7
|
||
price_out = 2.8
|
||
|
||
[[models]]
|
||
model_identifier = "Qwen/Qwen2.5-VL-72B-Instruct"
|
||
name = "qwen2.5-vl-72b"
|
||
api_provider = "SiliconFlow"
|
||
model_flags = [ "vision", "text",]
|
||
price_in = 4.13
|
||
price_out = 4.13
|
||
|
||
[[models]]
|
||
model_identifier = "FunAudioLLM/SenseVoiceSmall"
|
||
name = "sensevoice-small"
|
||
api_provider = "SiliconFlow"
|
||
model_flags = [ "audio",]
|
||
price_in = 0
|
||
price_out = 0
|
||
|
||
[[models]]
|
||
model_identifier = "BAAI/bge-m3"
|
||
name = "bge-m3"
|
||
api_provider = "SiliconFlow"
|
||
model_flags = [ "text", "embedding",]
|
||
price_in = 0
|
||
price_out = 0
|
||
|
||
|
||
[task_model_usage]
|
||
#llm_reasoning = {model="deepseek-r1", temperature=0.8, max_tokens=1024, max_retry=0}
|
||
#llm_normal = {model="deepseek-r1", max_tokens=1024, max_retry=0}
|
||
#embedding = "siliconflow-bge-m3"
|
||
#schedule = [
|
||
# "deepseek-v3",
|
||
# "deepseek-r1",
|
||
#] |