# 文本对话 #### 功能介绍 基于文本对话功能，用户可以与模型进行自然而流畅的对话和交流。 #### 授权信息 账号具备所有API的调用权限，如果使用账号下的IAM用户调用当前API，该IAM用户需具备调用API所需的权限，具体权限要求请参见[权限和授权项](https://support.huaweicloud.com/productdesc-pangulm/pangulm_01_0020.html)。 #### URI 获取URI方式请参见[请求URI](https://support.huaweicloud.com/api-pangulm/pangulm_05_0004.html)。 NLP推理服务提供两种推理接口调用： - 盘古推理接口（V1推理接口）。 - 业界通用的OpenAI格式接口（V2推理接口）。 两种接口定义如[表1]所示。 表1NLP服务推理接口 | API分类 | API访问路径（URI） | |:---|:---| | V1推理接口 | POST /v1/{project_id}/deployments/{deployment_id}/chat/completions | | V2推理接口 | POST /api/v2/chat/completions | V1推理接口URI需要输入额外参数，参数说明如[表2 V1推理接口路径参数]所示： 表2V1推理接口路径参数 | 参数 | 是否必选 | 参数类型 | 描述 | |:---|:---|:---|:---| | project_id | 是 | String | **参数解释：** 项目ID，获取方法请参见[获取项目ID](https://support.huaweicloud.com/api-pangulm/pangulm_05_0021.html)。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | deployment_id | 是 | String | **参数解释：** 模型的部署ID，获取方法请参见[获取模型部署ID](https://support.huaweicloud.com/api-pangulm/pangulm_05_0023.html)。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | #### 请求参数 V1、V2推理接口的鉴权方式不同，请求参数与响应参数也有不同，说明如下： **Header参数** 1. V1接口支持Token鉴权方式，也支持API Key鉴权方式。两种鉴权方式请求Header参数说明如下： - 使用[Token认证](https://support.huaweicloud.com/api-pangulm/pangulm_05_0009.html#pangulm_05_0009__section94702116509)方式的请求Header参数见[表3]。 表3请求Header参数（Token认证） | 参数 | 是否必选 | 参数类型 | 描述 | |:---|:---|:---|:---| | X-Auth-Token | 是 | String | **参数解释：** 用户Token。 用于获取操作API的权限。如[图4](https://support.huaweicloud.com/api-pangulm/pangulm_05_0009.html#pangulm_05_0009__fig10797003103)中响应消息头中X-Subject-Token的值即为Token。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | Content-Type | 是 | String | **参数解释：** 发送的实体的MIME类型。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** application/json | - 使用[API Key认证](https://support.huaweicloud.com/api-pangulm/pangulm_05_0009.html#pangulm_05_0009__section148602501261)方式的请求Header参数见[表4]。 表4请求Header参数（API Key认证） | 参数 | 是否必选 | 参数类型 | 描述 | |:---|:---|:---|:---| | X-Apig-AppCode | 是 | String | **参数解释：** API Key值。 用于获取操作API的权限。[API Key认证](https://support.huaweicloud.com/api-pangulm/pangulm_05_0009.html#pangulm_05_0009__section148602501261)响应消息头中X-Apig-AppCode的值即为API Key。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | Content-Type | 是 | String | **参数解释：** 发送的实体的MIME类型。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** application/json | 2. V2接口仅支持API Key鉴权方式。请求Header参数见[表5] 表5请求Header参数（OpenAI格式的API Key认证） | 参数 | 是否必选 | 参数类型 | 描述 | |:---|:---|:---|:---| | Authorization | 是 | String | **参数解释：** 用户创建应用接入获取的API Key，拼接"Bearer "后的字符串。示例：Bearer d59\*\*\*\*\*\*9C3。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | Content-Type | 是 | String | **参数解释：** 发送的实体的MIME类型。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** application/json | **请求** **Body参数** V1、V2推理接口请求Body参数一致，如[表6 请求Body参数]描述。 表6请求Body参数 | 参数 | 是否必选 | 参数类型 | 描述 | |:---|:---|:---|:---| | messages | 是 | Array of [message] objects | **参数解释：** 多轮对话问答对，包含两个属性：role和content。 - role表示对话的角色，取值是system或user。 如果需要模型以某个人设形象回答问题，可以将role参数设置为system。不使用人设时，可设置为user。在一次会话请求中，人设只需要设置一次。 - content表示对话的内容，可以是任意文本。 messages参数可以帮助模型根据对话的上下文生成合适的回复。 **约束限制：** 数组长度：**1 - 20** **取值范围：** 不涉及 **默认取值：** 不涉及 | | model | V1推理接口：否 V2推理接口：是 | String | **参数解释：** 使用的推理服务模型名称，为推理服务部署时指定的Deployed_Model，可在推理服务详情页面查询到。V2推理接口必须指定此参数，V1推理接口不需要此参数。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | stream | 否 | boolean | **参数解释：** 流式开关。 开启流式开关后，API会在生成文本的过程中，实时地将生成的文本发送给客户端，而不是等到生成完成后一次性将所有文本发送给客户端。 **约束限制：** 如果开启流式，请赋值true，同时n参数只能设置为1。 **取值范围：** 不涉及 **默认取值：** false | | temperature | 否 | Float | **参数解释：** 用于控制生成文本的多样性和创造力。 参数的取值范围是 (0, 1\]，取值接近0表示最低的随机性，1表示最高的随机性。一般来说，temperature越低，适合完成确定性的任务。temperature越高，例如0.9，适合完成创造性的任务。 temperature参数可以影响语言模型输出的质量和多样性，但也不是唯一的因素。还有其他一些参数，如top_p参数也可以用来调整语言模型的行为和偏好。 **约束限制：** 不涉及 **取值范围：** (0, 1\] **默认取值：** 不涉及 | | top_p | 否 | Float | **参数解释：** 一种替代温度采样的方法，称为nucleus sampling，其中模型考虑具有top_p概率质量的标记的结果。 **约束限制：** 不涉及 **取值范围：** (0, 1\] **默认取值：** 不涉及 | | max_tokens | 否 | Integer | **参数解释：** 生成文本的最大token数量，单位token。 说明： token是指模型处理和生成文本的基本单位。token可以是词或者字符的片段。模型的输入和输出的文本都会被转换成token，然后根据模型的概率分布进行采样或者计算。 **约束限制：** 输入的文本加上生成的文本总量不能超过模型所能处理的最大长度。 **取值范围：** 最小值：**1** 最大值：模型支持的最大token数 **默认取值：**模型支持的最大token数 | | presence_penalty | 否 | Float | **参数解释：** 用于调整模型对新Token的处理方式。即如果一个Token已经在之前的文本中出现过，那么模型在生成这个Token时会受到一定的惩罚。当presence_penalty的值为正数时，模型会更倾向于生成新的、未出现过的Token，即模型会更倾向于谈论新的话题。 **约束限制：** 不涉及 **取值范围：** 最小值：**-2** 最大值：**2** **默认取值：** **0 （表示该参数未生效）** | | frequency_penalty | 否 | Float | **参数解释：** 用于调整模型对频繁出现的Token的处理方式。即如果一个Token在训练集中出现的频率较高，那么模型在生成这个Token时会受到一定的惩罚。当frequency_penalty的值为正数时，模型会更倾向于生成出现频率较低的Token，即模型会更倾向于使用不常见的词汇。 **约束限制：** 不涉及 **取值范围：** 最小值：**-2** 最大值：**2** **默认取值：** **0 （表示该参数未生效）** | | stop | 否 | Union\[String, List\[String\]\] | **参数解释：** 停止符，控制模型输出何时结束的参数。可以指定一个或多个字符串作为"停止符"。当模型在生成过程中产生了任何一个停止符时，它会立即停止生成更多内容。停止符本身不会出现在最终输出中。例如：\["你"，"好"\]，生成文本时遇到"你"或者"好"将停止文本生成。 **约束限制：** 非空字符串或字符串列表。 **取值范围：** 不涉及 **默认取值：** 不涉及 | | moderation_config | 否 | [moderation_config] objects | **参数解释：** 盘古护栏审核配置。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | enable_search | 否 | boolean | **参数解释：** 是否启动搜索增强。 **约束限制：** 不涉及 **取值范围：** - **false**：关闭 - **true**：启动 **默认取值：** 不涉及 | 表7moderation_config | 参数 | 参数类型 | 描述 | |:---|:---|:---| | black_glossary_names | List\ | **参数解释：** 黑名单词库列表。 **约束限制：** 仅对开启高级版内容审核的推理服务生效。 **取值范围：** 不涉及 **默认取值：** 不涉及 | | white_glossary_names | List\ | **参数解释：** 白名单词库列表。 **约束限制：** 仅对开启高级版内容审核的推理服务生效。 **取值范围：** 不涉及 **默认取值：** 不涉及 | | question_moderation | boolean | **参数解释：** 是否开启对提示词进行内容审核，仅对订购用户生效。 **约束限制：** 不涉及 **取值范围：** - **true**：审核 - **false**：不审核 **默认取值：** **true** | | answer_moderation | boolean | **参数解释：** 是否开启对推理结果进行内容审核，仅对订购用户生效。 **约束限制：** 不涉及 **取值范围：** - **true**：审核 - **false**：不审核 **默认取值：** **true** | | show_result | boolean | **参数解释：** 是否返回内容审核不通过原因 缺省值：**false** **约束限制：** 不涉及 **取值范围：** **true**：返回 **false**：不返回 **默认取值：** 不涉及 | 表8message | 参数 | 是否必选 | 参数类型 | 描述 | |:---|:---|:---|:---| | role | V1推理接口：否 V2推理接口：是 | String | **参数解释：** 对话的角色。 如果需要模型以某个人设形象回答问题，可以将role参数设置为system。不使用人设时，可设置为user。在一次会话请求中，人设只需要设置一次。多轮对话中，用户输入提示词的role设置为user，推理结果的role设置为assistant。 **约束限制：** 不涉及 **取值范围：** system、user、assistant **默认取值：** 不涉及 | | content | 是 | String | **参数解释：** 对话的内容，可以是任意文本，单位token。 **约束限制：** 设置多轮对话时，message中content个数不能超过20。 **取值范围：** 最小长度：**1** 最大长度：N1模型支持32K/128K，N2模型支持32K/128K，N4模型支持4K/32K。 **默认取值：** 不涉及 | #### 响应参数 **非流式** **状态码： 200** 表9响应Body参数 | 参数 | 参数类型 | 描述 | |:---|:---|:---| | id | String | **参数解释：** 用来标识每个响应的唯一字符串。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | object | String | **参数解释：** 固定为"chat.completion"。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** chat.completion | | created | Integer | **参数解释：** 响应生成的时间。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | model | String | **参数解释：** 请求模型名称标识。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | choices | Array of [choices] objects | **参数解释：** 生成的文本列表。包含以下属性： - message：生成的文本内容。 - index：生成的文本在列表中的索引，从0开始。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | usage | [usage] object | **参数解释：** 该参数可以帮助用户了解和控制模型的使用情况，避免超出Tokens限制。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | 表10choices | 参数 | 参数类型 | 描述 | |:---|:---|:---| | message | [message] object | **参数解释：** 生成的文本内容。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | index | Integer | **参数解释：** 生成的文本在列表中的索引，从0开始。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | finish_reason | String | **参数解释：** 模型停止生成token的原因。 **约束限制：** 不涉及 **取值范围：** \[stop, length, content_filter, tool_calls, insufficient_system_resource\] - stop：模型自然停止生成，或遇到stop序列中列出的字符串。 - length ：输出长度达到了模型上下文长度限制，或达到了max_tokens的限制。 - content_filter：输出内容因触发过滤策略而被过滤。 - tool_calls：模型决定调用外部工具（函数/API）来完成任务。 - insufficient_system_resource：系统推理资源不足，生成被打断。 **默认取值：** **stop** | | stop_reason | Union\[Integer, String\] | **参数解释：** 导致生成停止的token id或者字符串。如果是遇到EOS token则返回默认值。如果是因为用户请求参数中指定的stop参数中的字符串或者token id，则返回对应的字符串或者token id。不是openai接口标准字段，但vllm接口支持。 缺省值：**None** **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | 表11message | 参数 | 参数类型 | 描述 | |:---|:---|:---| | role | String | **参数解释：** 对话的角色。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | content | String | **参数解释：** 对话的内容。 **约束限制：** 不涉及 **取值范围：** 最小长度：**1** 最大长度：不同模型支持的token长度请参见《产品介绍》"模型能力与规格"章节。 **默认取值：** 不涉及 | | reasoning_content | String | **参数解释：** 大模型推理过程中在给出最终答案之前的推理内容。 **约束限制：** 该字段仅适用于Pangu-NLP-Reasoner系列模型。 **取值范围：** 不涉及 **默认取值：** 不涉及 | 表12usage | 参数 | 参数类型 | 描述 | |:---|:---|:---| | completion_tokens | Number | **参数解释：** 推理服务返回结果的Token数。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | prompt_tokens | Number | **参数解释：** 用户输入的提示词及默认人设的Token数。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | total_tokens | Number | **参数解释：** 总消耗Token数。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | **流式（stream参数为true）** **状态码： 200** 表13流式输出的数据单元 | 参数 | 参数类型 | 描述 | |:---|:---|:---| | data | [CompletionStreamResponse] | **参数解释：** stream=true时，模型生成的消息以流式形式返回。生成的内容以增量的方式逐步发送回来，每个data字段均包含一部分生成的内容，直到所有data返回，响应结束。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | 表14CompletionStreamResponse | 参数 | 参数类型 | 描述 | |:---|:---|:---| | id | String | **参数解释：** 该对话的唯一标识符。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | created | Integer | **参数解释：** 创建聊天完成时的Unix时间戳（以秒为单位）。流式响应的每个chunk的时间戳相同。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | model | String | **参数解释：** 生成该completion的模型名。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | object | String | **参数解释：** 对象的类型，其值为chat.completion.chunk。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** chat.completion.chunk | | choices | [ChatCompletionResponseStreamChoice] | **参数解释：** 模型生成的completion的选择列表。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | usage | [UsageInfo] | **参数解释：** 该对话请求的Token用量信息。该参数可以帮助用户了解和控制模型的使用情况，避免超出Tokens限制。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | 表15ChatCompletionResponseStreamChoice | 参数 | 参数类型 | 描述 | |:---|:---|:---| | index | Integer | **参数解释：** 该completion在模型生成的completion的选择列表中的索引。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | finish_reason | String | **参数解释：** 模型停止生成token的原因。 **约束限制：** 不涉及 **取值范围：** \[stop, length, content_filter, tool_calls, insufficient_system_resource\] - stop：模型自然停止生成，或遇到stop序列中列出的字符串。 - length ：输出长度达到了模型上下文长度限制，或达到了max_tokens的限制。 - content_filter：输出内容因触发过滤策略而被过滤。 - tool_calls：模型决定调用外部工具（函数/API）来完成任务。 - insufficient_system_resource：系统推理资源不足，生成被打断。 **默认取值：** 不涉及 | | delta | [DeltaMessage] | **参数解释：** V2推理接口流式返回的一个completion增量。 **约束限制：** V1推理接口返回体不包含此参数。 **取值范围：** 不涉及 **默认取值：** 不涉及 | | message | [DeltaMessage] | **参数解释：** V1推理接口流式返回的一个completion增量。 **约束限制：** V2推理接口返回体不包含此参数。 **取值范围：** 不涉及 **默认取值：** 不涉及 | 表16DeltaMessage | 参数 | 参数类型 | 描述 | |:---|:---|:---| | role | String | **参数解释：** 产生这条消息的角色。 **约束限制：** 不涉及 **取值范围：** assistant **默认取值：** assistant | | content | String | **参数解释：** completion增量的内容。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | reasoning_content | String | **参数解释：** 大模型推理过程中在给出最终答案之前的推理内容（模型的思考过程）。 **约束限制：** 仅适用于Pangu-NLP-Reasoner系列模型。 **取值范围：** 不涉及 **默认取值：** 不涉及 | 表17UsageInfo | 参数 | 参数类型 | 描述 | |:---|:---|:---| | prompt_tokens | Integer | **参数解释：** 用户输入的提示词及默认人设的Token数。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | completion_tokens | Integer | **参数解释：** 推理服务返回结果的Token数。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | total_tokens | Integer | **参数解释：** 总消耗Token数。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | 表18内容审核拦截场景，流式输出的数据单元 | 参数 | 参数类型 | 描述 | |:---|:---|:---| | suggestion | String | **参数解释：** 审核结果：block表示未通过。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | reply | String | **参数解释：** 兜底回复：审核未通过时兜底回复为有效回复，兜底策略。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | **状态码： 400** 接口报错的场景下，V1推理接口返回的报错信息符合华为云规范；V2推理接口则会对外透传推理服务返回的错误信息，通常符合OpenAI接口格式。 表19V1推理接口响应错误信息Body参数 | 参数 | 参数类型 | 描述 | |:---|:---|:---| | error_msg | String | **参数解释：** 错误信息。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | error_code | String | **参数解释：** 错误码。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | details | List\ | **参数解释：** 推理服务返回的报错信息，具体的格式、内容取决于推理服务。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | 表20V2推理接口响应错误信息Body参数 | 参数 | 参数类型 | 描述 | |:---|:---|:---| | error | ErrorResp | **参数解释：** 错误信息。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | id | String | **参数解释：** 请求ID。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | 表21ErrorResp | 参数 | 参数类型 | 描述 | |:---|:---|:---| | code | String | **参数解释：** 错误码。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | type | String | **参数解释：** 错误类型。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | | message | String | **参数解释：** 错误详情。 **约束限制：** 不涉及 **取值范围：** 不涉及 **默认取值：** 不涉及 | #### 请求示例 - 单轮问答 ``` V1推理接口： POST https://{endpoint}/v1/{project_id}/deployments/{deployment_id}/chat/completions Request Header: Content-Type: application/json X-Auth-Token: MIINRwYJKoZIhvcNAQcCoIINODCCDTQCAQExDTALBglghkgBZQMEAgEwgguVBgkqhkiG... Request Body: { "model": "pangu-nlp-n1-32k", // 仅V2接口需要此参数 "messages": [ { "role":"user", "content": "介绍下长江，以及长江中典型的鱼类" } ], "temperature": 0.9, "max_tokens": 600 } ``` ``` V2推理接口： POST https://{endpoint}/api/v2/chat/completions Request Header: Content-Type: application/json Authorization: Bearer 201ca68f-45f9-4e19-8fa4-831e... Request Body: { "model": "pangu-nlp-n1-32k", // 仅V2接口需要此参数 "messages": [ { "role":"user", "content": "介绍下长江，以及长江中典型的鱼类" } ], "temperature": 0.9, "max_tokens": 600 } ``` - 单轮流式问答（stream参数值为true） ``` { "model": "pangu-nlp-n1-32k", // 仅V2接口需要此参数 "messages": [ { "role":"user", "content": "五岳分别是哪些山" } ], "temperature": 0.9, "max_tokens": 600, "stream": "true" } ``` - 带有人设的单轮问答（role参数值为system） ``` { "model": "pangu-nlp-n1-32k", // 仅V2接口需要此参数 "messages": [ { "role": "system", "content": "你的名字叫智子，是一名幼儿园老师，请用幼儿园老师的口吻回答问题，注意语气温和亲切，通过提问、引导、赞美等方式，激发学生的思维和想象力。" }, { "role": "user", "content": "介绍下长江，以及长江中典型的鱼类" } ], "temperature": 0.9, "max_tokens": 600 } ``` - 带有人设的单轮流式问答（role参数值为system，stream参数值为true） ``` { "model": "pangu-nlp-n1-32k", // 仅V2接口需要此参数 "messages": [ { "role": "system", "content": "你的名字叫智子，是一名幼儿园老师，请用幼儿园老师的口吻回答问题，注意语气温和亲切，通过提问、引导、赞美等方式，激发学生的思维和想象力。" }, { "role": "user", "content": "写一首诗" } ], "temperature": 0.9, "max_tokens": 600, "stream": "true" } ``` - 多轮问答 ``` # 多轮问答传参方法：在完成第一轮问答，进行第二轮问答时，需要将第一轮的问题和答案、第二轮问题作为参数传入……依次类推，完成多轮对话。 { "model": "pangu-nlp-n1-32k", // 仅V2接口需要此参数 "messages": [ { "role":"user", "content": "介绍下长江，以及长江中典型的鱼类" //第一轮问题 }, { "role":"assistant", "content": "长江是中国第一大河,全长6300多公里。它发源于唐古拉山脉南麓,流经中国11个省市自治区和两个特别行政区,最终注入东海。\n长江中典型的鱼类有:鲢鱼、鳙鱼、草鱼、鲤鱼等。" //第一轮答案 }, { "role":"user", "content": "途径的省份列2个，并说明有哪些人文风景" //第二轮问题 }, { "role":"assistant", "content": "1. 湖南 - 长沙岳麓山、张家界天门山等\n2. 四川 - 九寨沟、峨眉山等" //第二轮答案 }, { "role":"user", "content": "提到的湖南景点，详细说明下" //第三轮问题 } ], "temperature": 0.9, "max_tokens": 600 } ``` - 多轮流式问答（stream参数值为true） ``` # 多轮问答传参方法：在完成第一轮问答，进行第二轮问答时，需要将第一轮的问题和答案、第二轮问题作为参数传入……依次类推，完成多轮对话。 { "model": "pangu-nlp-n1-32k", // 仅V2接口需要此参数 "messages": [ { "role":"user", "content": "介绍下长江，以及长江中典型的鱼类" //第一轮问题 }, { "role":"assistant", "content": "长江是中国第一大河,全长6300多公里。它发源于唐古拉山脉南麓,流经中国11个省市自治区和两个特别行政区,最终注入东海。\n长江中典型的鱼类有:鲢鱼、鳙鱼、草鱼、鲤鱼等。" //第一轮答案 }, { "role":"user", "content": "途径的省份列2个，并说明有哪些人文风景" //第二轮问题 }, { "role":"assistant", "content": "1. 湖南 - 长沙岳麓山、张家界天门山等\n2. 四川 - 九寨沟、峨眉山等" //第二轮答案 }, { "role":"user", "content": "张家界具体在哪里" //第三轮问题 } ], "temperature": 0.9, "max_tokens": 600, "stream": "true" } ``` - 带有人设的多轮流式问答（role参数值为system，stream参数值为true） ``` { "model": "pangu-nlp-n1-32k", // 仅V2接口需要此参数 "messages": [ { "content": "你是一个热心的导游", //人设 "role": "system" }, { "role":"user", "content": "介绍下长江，以及长江中典型的鱼类" //第一轮问题 }, { "role":"assistant", "content": "长江，亚洲最长河流，世界第三长河流，它发源于唐古拉山脉南麓，流经我国多个省份，最终注入东海。长江流域资源丰富，特别是鱼类资源，例如：鲥鱼、白鲟、白鳍豚等，都是长江特有的珍稀物种。" //第一轮答案 }, { "role":"user", "content": "途径的省份列2个，并说明有哪些人文风景" //第二轮问题 }, { "role":"assistant", "content": "长江流经的省份众多，这里为您介绍两个：湖北和四川。湖北，长江流经此省9年，有武汉、宜昌等知名城市，人文风景丰富，如黄鹤楼、古琴台等。四川，长江流经此省14年，有成都、重庆等大城市，人文风景包括都江堰、乐山大佛等世界文化遗产。" //第二轮答案 }, { "role":"user", "content": "张家界具体在哪里" //第三轮问题 } ], "temperature": 0.5, "max_tokens": 800, "stream": true } ``` #### 响应示例 **状态码： 200** OK - 非流式问答响应 ``` { "id": "chat-3f7d8e038f244ae5b7cb0cecf352e154", "object": "chat.completion", "created": 1745725506, "model": "pangu-nlp-n1-32k_kpyq10", "choices": [ { "index": 0, "message": { "role": "assistant", "content": "你好！有什么我可以帮助你的吗？", "tool_calls": [] }, "logprobs": null, "finish_reason": "stop", "stop_reason": null } ], "usage": { "prompt_tokens": 64, "total_tokens": 73, "completion_tokens": 9 }, "prompt_logprobs": null } ``` - 带有思维链的非流式问答响应 ``` { "id": "4c0dcef2-7f8f-4c57-93c7-484c03b0a216", "object": "chat.completion", "created": 1741937420, "model": "pangu-nlp-n1-32k_kpyq10", "choices": [ { "index": 0, "message": { "role": "assistant", "content": "\n\n你好！很高兴见到你，有什么我可以帮忙的吗？", "reasoning_content": "嗯，用户发来“你好”，这是一个常见的中文问候。我要用中文回应用户，保持友好和亲切。首先要分析用户的需求，可能只是想打个招呼，或者有具体的问题想问。\n\n接下来要考虑用户可能的背景，可能来自不同的地方，但使用简体中文，所以可能是中国大陆的用户。需要保持回答的通用性，适合所有使用中文的人。\n\n还要判断是否有更深层次的需求，用户可能在测试聊天机器人，或者真的需要帮助。这时候需要给出一个开放式的回应，鼓励用户进一步说明需求。例如询问有什么可以帮忙的，这样既回应了问候，又引导用户提出具体问题。\n\n另外要注意语言的流畅和自然，避免过于机械化的回答。保持礼貌和专业，同时适当使用表情符号增加亲和力，比如用[微笑]或者之类的，不过可能根据平台的不同需要调整。\n\n最后要确保回答符合规范，不涉及敏感话题，内容积极健康。检查是否有拼写或语法错误，确保回复准确无误。\n", "tool_calls": [] }, "logprobs": null, "finish_reason": "stop", "stop_reason": null } ], "usage": { "prompt_tokens": 22, "total_tokens": 377, "completion_tokens": 355 }, "prompt_logprobs": null } ``` - 流式问答响应 ``` V1推理接口返回体： data: {"id":"chat-59170add0fd1427bbca0388431058d45","object":"chat.completion.chunk","created":1745725837,"model":"pangu-nlp-n1-32k_kpyq10","choices":[{"index":0,"logprobs":null,"finish_reason":null,"message":{"role":"assistant"}}],"usage":{"prompt_tokens":64,"total_tokens":64,"completion_tokens":0}} data: {"id":"chat-59170add0fd1427bbca0388431058d45","object":"chat.completion.chunk","created":1745725837,"model":"pangu-nlp-n1-32k_kpyq10","choices":[{"index":0,"logprobs":null,"finish_reason":null,"message":{"content":"你好"}}],"usage":{"prompt_tokens":64,"total_tokens":65,"completion_tokens":1}} data: {"id":"chat-59170add0fd1427bbca0388431058d45","object":"chat.completion.chunk","created":1745725837,"model":"pangu-nlp-n1-32k_kpyq10","choices":[{"index":0,"logprobs":null,"finish_reason":"stop","stop_reason":null,"message":{"content":"！有什么我可以帮你的吗？"}}],"usage":{"prompt_tokens":64,"total_tokens":73,"completion_tokens":9}} data: {"id":"chat-59170add0fd1427bbca0388431058d45","object":"chat.completion.chunk","created":1745725837,"model":"pangu-nlp-n1-32k_kpyq10","choices":[],"usage":{"prompt_tokens":64,"total_tokens":73,"completion_tokens":9}} event:{"usage":{"completionTokens":9,"promptTokens":64,"totalTokens":73},"tokens":64,"token_number":9} data:[DONE] ``` ``` V2推理接口返回体： data: {"id":"chat-b9417f06b6524362ae09844cc9b0172d","object":"chat.completion.chunk","created":1745725924,"model":"pangu-nlp-n1-32k_kpyq10","choices":[{"index":0,"delta":{"role":"assistant"},"logprobs":null,"finish_reason":null}],"usage":{"prompt_tokens":64,"total_tokens":64,"completion_tokens":0}} data: {"id":"chat-b9417f06b6524362ae09844cc9b0172d","object":"chat.completion.chunk","created":1745725924,"model":"pangu-nlp-n1-32k_kpyq10","choices":[{"index":0,"delta":{"content":"你好"},"logprobs":null,"finish_reason":null}],"usage":{"prompt_tokens":64,"total_tokens":65,"completion_tokens":1}} data: {"id":"chat-b9417f06b6524362ae09844cc9b0172d","object":"chat.completion.chunk","created":1745725924,"model":"pangu-nlp-n1-32k_kpyq10","choices":[{"index":0,"delta":{"content":"！有什么我可以帮助你的吗？"},"logprobs":null,"finish_reason":"stop","stop_reason":null}],"usage":{"prompt_tokens":64,"total_tokens":73,"completion_tokens":9}} data: {"id":"chat-b9417f06b6524362ae09844cc9b0172d","object":"chat.completion.chunk","created":1745725924,"model":"pangu-nlp-n1-32k_kpyq10","choices":[],"usage":{"prompt_tokens":64,"total_tokens":73,"completion_tokens":9}} data:[DONE] ``` - 带有思维链的流式问答响应 ``` V1推理接口返回体： data: {"id":"chat-cfc9f0e8404540af97aa97bfc5c5090f","object":"chat.completion.chunk","created":1747364482,"model":"pangu-nlp-n2-reasoner-32k","choices":[{"index":0,"message":{"role":"assistant"},"logprobs":null,"finish_reason":null}],"usage":{"prompt_tokens":23,"total_tokens":23,"completion_tokens":0}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": ""}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 24, "completion_tokens": 1}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": ""}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 25, "completion_tokens": 2}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": ""}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 26, "completion_tokens": 3}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": "好的"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 27, "completion_tokens": 4}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": "，"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 28, "completion_tokens": 5}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": "用户"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 29, "completion_tokens": 6}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": "发"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 30, "completion_tokens": 7}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": "来了"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 31, "completion_tokens": 8}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": "“"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 32, "completion_tokens": 9}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": "你好"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 33, "completion_tokens": 10}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": "”"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 34, "completion_tokens": 11}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": "，"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 35, "completion_tokens": 12}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": "我需要"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 36, "completion_tokens": 13}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": "回复"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 37, "completion_tokens": 14}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": "他们"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 38, "completion_tokens": 15}} ... data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": "符合"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 178, "completion_tokens": 155}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": "要求"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 179, "completion_tokens": 156}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": "。"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 180, "completion_tokens": 157}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": "\n"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 181, "completion_tokens": 158}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": ""}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 182, "completion_tokens": 159}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"reasoning_content": ""}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 183, "completion_tokens": 160}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": "\n\n"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 184, "completion_tokens": 161}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": "你好"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 185, "completion_tokens": 162}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": "！"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 186, "completion_tokens": 163}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": "我是"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 187, "completion_tokens": 164}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": "华为"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 188, "completion_tokens": 165}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": "云"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 189, "completion_tokens": 166}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": "开发的"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 190, "completion_tokens": 167}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": "盘"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 191, "completion_tokens": 168}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": "古"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 192, "completion_tokens": 169}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": "大"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 193, "completion_tokens": 170}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": "语言"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 194, "completion_tokens": 171}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": "模型"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 195, "completion_tokens": 172}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": "。"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 196, "completion_tokens": 173}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": "有什么"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 197, "completion_tokens": 174}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": "可以"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 198, "completion_tokens": 175}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": "帮"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 199, "completion_tokens": 176}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": "您的"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 200, "completion_tokens": 177}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": "？"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 201, "completion_tokens": 178}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "message": {"content": ""}, "logprobs": null, "finish_reason": "stop", "stop_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 202, "completion_tokens": 179}} data: {"id":"chat-cfc9f0e8404540af97aa97bfc5c5090f","object":"chat.completion.chunk","created":1747364482,"model":"pangu-nlp-n2-reasoner-32k","choices":[],"usage":{"prompt_tokens":23,"total_tokens":202,"completion_tokens":179}} data:[DONE] ``` ``` V2推理接口返回体： data: {"id":"chat-cfc9f0e8404540af97aa97bfc5c5090f","object":"chat.completion.chunk","created":1747364482,"model":"pangu-nlp-n2-reasoner-32k","choices":[{"index":0,"delta":{"role":"assistant"},"logprobs":null,"finish_reason":null}],"usage":{"prompt_tokens":23,"total_tokens":23,"completion_tokens":0}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": ""}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 24, "completion_tokens": 1}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": ""}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 25, "completion_tokens": 2}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": ""}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 26, "completion_tokens": 3}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": "好的"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 27, "completion_tokens": 4}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": "，"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 28, "completion_tokens": 5}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": "用户"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 29, "completion_tokens": 6}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": "发"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 30, "completion_tokens": 7}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": "来了"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 31, "completion_tokens": 8}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": "“"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 32, "completion_tokens": 9}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": "你好"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 33, "completion_tokens": 10}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": "”"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 34, "completion_tokens": 11}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": "，"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 35, "completion_tokens": 12}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": "我需要"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 36, "completion_tokens": 13}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": "回复"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 37, "completion_tokens": 14}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": "他们"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 38, "completion_tokens": 15}} ... data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": "符合"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 178, "completion_tokens": 155}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": "要求"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 179, "completion_tokens": 156}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": "。"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 180, "completion_tokens": 157}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": "\n"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 181, "completion_tokens": 158}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": ""}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 182, "completion_tokens": 159}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"reasoning_content": ""}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 183, "completion_tokens": 160}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": "\n\n"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 184, "completion_tokens": 161}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": "你好"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 185, "completion_tokens": 162}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": "！"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 186, "completion_tokens": 163}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": "我是"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 187, "completion_tokens": 164}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": "华为"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 188, "completion_tokens": 165}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": "云"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 189, "completion_tokens": 166}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": "开发的"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 190, "completion_tokens": 167}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": "盘"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 191, "completion_tokens": 168}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": "古"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 192, "completion_tokens": 169}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": "大"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 193, "completion_tokens": 170}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": "语言"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 194, "completion_tokens": 171}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": "模型"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 195, "completion_tokens": 172}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": "。"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 196, "completion_tokens": 173}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": "有什么"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 197, "completion_tokens": 174}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": "可以"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 198, "completion_tokens": 175}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": "帮"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 199, "completion_tokens": 176}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": "您的"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 200, "completion_tokens": 177}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": "？"}, "logprobs": null, "finish_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 201, "completion_tokens": 178}} data: {"id": "chat-cfc9f0e8404540af97aa97bfc5c5090f", "object": "chat.completion.chunk", "created": 1747364482, "model": "pangu-nlp-n2-reasoner-32k", "choices": [{"index": 0, "delta": {"content": ""}, "logprobs": null, "finish_reason": "stop", "stop_reason": null}], "usage": {"prompt_tokens": 23, "total_tokens": 202, "completion_tokens": 179}} data: {"id":"chat-cfc9f0e8404540af97aa97bfc5c5090f","object":"chat.completion.chunk","created":1747364482,"model":"pangu-nlp-n2-reasoner-32k","choices":[],"usage":{"prompt_tokens":23,"total_tokens":202,"completion_tokens":179}} data:[DONE] ``` #### 状态码 请参见[状态码](https://support.huaweicloud.com/api-pangulm/pangulm_05_0019.html)。 #### 错误码 请参见[错误码](https://support.huaweicloud.com/api-pangulm/pangulm_05_0020.html)。