diff --git a/cmd/testdata/aiconfig/ai_config.example.yaml b/cmd/testdata/aiconfig/ai_config.example.yaml index eea0fe8..12cdf98 100644 --- a/cmd/testdata/aiconfig/ai_config.example.yaml +++ b/cmd/testdata/aiconfig/ai_config.example.yaml @@ -1,33 +1,33 @@ -# 使用的时候,将文件复制到 cmd/testdata/aiconfig/ 目录下,并重命名为 ai_config.yaml +# To use this file, copy it to cmd/testdata/aiconfig/ and rename it to ai_config.yaml. # -# 配置规则说明: -# 1. name: 配置的唯一标识,系统通过此字段进行去重和更新 -# 2. provider: 模型提供商,固定写:openai,系统目前ai接口调用都走的openai协议 -# 3. baseUrl: API 服务地址 -# 4. apiKey: API 密钥 -# 5. modelType: 模型类型,必须为以下之一: -# - llm: 大语言模型,用于对话和推理 -# - embedding: 文本向量化模型,用于生成文本嵌入向量 -# - rerank: 排序模型,用于重排序搜索结果 -# 6. modelName: 具体的模型名称,由提供商指定 -# 7. dimension: 向量维度 -# - llm 和 rerank: 通常设为 0 -# - embedding: 向量维度数,如 OpenAI text-embedding-3-small 为 1536 -# 8. maxContextTokens: 最大上下文令牌数,0 表示无限制 -# - llm 和 embedding: 分别表示模型支持的最大输入令牌数 -# - rerank: 通常设为 0 -# 9. maxOutputTokens: 最大输出令牌数 -# - llm: 模型支持的最大输出令牌数 -# - embedding 和 rerank: 通常设为 0 -# 10. timeoutMs: 请求超时时间(毫秒) -# 11. maxRetryCount: 失败重试次数 -# 12. rpmLimit: 每分钟请求限制,0 表示无限制 -# 13. tpmLimit: 每分钟令牌限制,0 表示无限制 -# 15. sortNo: 排序号,用于同类型模型的优先级排序(数字越小优先级越高) -# 16. remark: 备注信息,用于描述此配置的用途或特殊说明 +# Configuration guide: +# 1. name: Unique identifier for the configuration. The system uses this field for deduplication and updates. +# 2. provider: Model provider. Always set this to openai because the current AI API calls use the OpenAI-compatible protocol. +# 3. baseUrl: API service endpoint. +# 4. apiKey: API key. +# 5. modelType: Model type. Must be one of the following: +# - llm: Large language model for chat and reasoning. +# - embedding: Text embedding model for generating vector representations of text. +# - rerank: Ranking model for reordering search results. +# 6. modelName: Specific model name, as defined by the provider. +# 7. dimension: Vector dimension. +# - llm and rerank: Usually set to 0. +# - embedding: Number of vector dimensions. For example, OpenAI text-embedding-3-small uses 1536. +# 8. maxContextTokens: Maximum context tokens. Use 0 for no limit. +# - llm and embedding: The maximum number of input tokens supported by the model. +# - rerank: Usually set to 0. +# 9. maxOutputTokens: Maximum output tokens. +# - llm: The maximum number of output tokens supported by the model. +# - embedding and rerank: Usually set to 0. +# 10. timeoutMs: Request timeout in milliseconds. +# 11. maxRetryCount: Number of retry attempts after a failed request. +# 12. rpmLimit: Requests per minute limit. Use 0 for no limit. +# 13. tpmLimit: Tokens per minute limit. Use 0 for no limit. +# 15. sortNo: Sort order for prioritizing models of the same type. Lower numbers have higher priority. +# 16. remark: Notes describing the purpose of this configuration or any special considerations. # items: - - name: LLM模型 + - name: LLM provider: openai baseUrl: https://dashscope.aliyuncs.com/compatible-mode/v1 apiKey: @@ -41,9 +41,9 @@ items: rpmLimit: 0 tpmLimit: 0 sortNo: 10 - remark: LLM模型 + remark: LLM - - name: 向量模型 + - name: embedding provider: openai baseUrl: https://dashscope.aliyuncs.com/compatible-mode/v1 apiKey: @@ -57,9 +57,9 @@ items: rpmLimit: 0 tpmLimit: 0 sortNo: 20 - remark: 向量模型 + remark: embedding - - name: 重排模型 + - name: rerank provider: openai baseUrl: https://api.cohere.com/v1 apiKey: @@ -73,4 +73,4 @@ items: rpmLimit: 0 tpmLimit: 0 sortNo: 30 - remark: 重排模型 + remark: rerank \ No newline at end of file