103 lines
3.3 KiB
YAML
103 lines
3.3 KiB
YAML
server:
|
||
ip: 0.0.0.0
|
||
port: 50055
|
||
accessToken: "me256487ang1chubdpdialoud22sev1ozhoguumyqca"
|
||
log:
|
||
# panic,fatal,error,warn,warning,info,debug,trace
|
||
level: "info"
|
||
logPath: "runtime/logs/app.log"
|
||
chat:
|
||
# openai key
|
||
api_key: "i0jey84SdkFdw5u43780yjr3h7se8nth0yi295nr94ksDngKprEh"
|
||
# openai 接口地址
|
||
base_url: "https://api.moonshot.cn/v1"
|
||
# 使用的训练模型
|
||
model: "gpt-3.5-turbo"
|
||
# 单次请求的上下文总长度,包括：请求消息+响应消息
|
||
max_tokens: 4096
|
||
# 表示语言模型输出的随机性和创造性
|
||
# 取值范围 0 ~ 1，值越大随机性越高
|
||
temperature: 0.8
|
||
# 用于生成文本时控制选词的随机程度
|
||
# 即下一个预测单词考虑的概率范围
|
||
# 取值范围0 ~ 1
|
||
top_p: 0.9
|
||
# 存在惩罚，用于生成文本时控制重复使用单词的程度
|
||
# 取值范围 0 ~ 1 ，0表示不惩罚，1表示禁止重复
|
||
presence_penalty: 0.8
|
||
# 用于控制模型生成回复时重复单词出现的频率
|
||
# 取值范围 0~1，值越大表示回复时会更注重避免使用已经出现的单词
|
||
frequency_penalty: 0.5
|
||
# AI助手特征描述
|
||
bot_desc: "你是一个AI助手，我需要你模拟一名资深的软件工程师来回答我的问题"
|
||
# 单次请求，保留的响应tokens数量
|
||
min_response_tokens: 2048
|
||
# 上下文缓存时长，单位s
|
||
context_ttl: 1800
|
||
# 上下文消息条数
|
||
context_len: 4
|
||
redis:
|
||
host: "127.0.0.1"
|
||
port: 8888
|
||
pwd: "123456"
|
||
mysql:
|
||
dsn: "root:root@tcp(127.0.0.1:3306)/ai_chat?collation=utf8mb4_unicode_ci&charset=utf8mb4"
|
||
maxLifeTime: 3600
|
||
maxOpenConn: 10
|
||
maxIdleConn: 10
|
||
dependOn:
|
||
sensitive:
|
||
address: "localhost:50053"
|
||
accessToken: "ang1chubdev1ozhome256487d22sapguuv1ozhom"
|
||
keywords:
|
||
address: "localhost:50054"
|
||
accessToken: "ang1chubdev1ozhome256487d22sapguuv1ozhom"
|
||
tokenizer:
|
||
address: "http://127.0.0.1:3002"
|
||
vector:
|
||
# 向量后端：tencent / pgvector
|
||
provider: "pgvector"
|
||
# 历史问答命中阈值
|
||
threshold: 0.99
|
||
tencent:
|
||
url: "http://lb-4u4r1fk4-1ys6gv3rpmdan420.clb.ap-guangzhou.tencentclb.com:60000"
|
||
username: "root"
|
||
pwd: "YaUfVueWZJ20e4ghyLlBT8Dou5OapwpFTUq50oft"
|
||
database: "ai-chat"
|
||
timeout: 5
|
||
maxIdleConnPerHost: 2
|
||
readConsistency: "eventualConsistency"
|
||
idleConnTimeout: 60
|
||
pgvector:
|
||
dsn: "postgres://postgres:postgres@127.0.0.1:15432/ai_chat?sslmode=disable"
|
||
table: "chat_record_vectors"
|
||
dimensions: 1024
|
||
maxLifeTime: 3600
|
||
maxOpenConn: 10
|
||
maxIdleConn: 10
|
||
embedding:
|
||
provider: "openai-compatible"
|
||
# 智谱 OpenAI 兼容网关；可被项目根目录 .env 覆盖
|
||
base_url: "https://open.bigmodel.cn/api/paas/v4"
|
||
# 默认故意设成错误值，真实 key 请放到项目根目录 .env
|
||
api_key: "__INVALID_SET_AI_CHAT_EMBEDDING_API_KEY__"
|
||
# embedding-2 固定 1024 维，和当前 pgvector 表结构一致
|
||
model: "embedding-2"
|
||
timeout: 10
|
||
vectorDB:
|
||
# 访问地址
|
||
url: "http://lb-4u4r1fk4-1ys6gv3rpmdan420.clb.ap-guangzhou.tencentclb.com:60000"
|
||
# 用户名
|
||
username: "root"
|
||
# 密码
|
||
pwd: "YaUfVueWZJ20e4ghyLlBT8Dou5OapwpFTUq50oft"
|
||
database: "ai-chat"
|
||
# 请求超时时长s
|
||
timeout: 5
|
||
# 最大空闲连接数
|
||
maxIdleConnPerHost: 2
|
||
# 读一致性： strongConsistency(强一致性)，eventualConsistency(最终一致性)
|
||
readConsistency: "eventualConsistency"
|
||
# 空闲连接超时时长s
|
||
idleConnTimeout: 60
|