This pull request includes several changes aimed at improving error handling, adding new functionalities, and refactoring existing code. The changes primarily focus on the AI provider and controller modules.
Error Handling Improvements:
Added checks to ensure the client is initialized before performing operations in multiple functions (PullModel, StopPull, CancelPipeline, RemoveModel, ModelsInstalled) in ai-provider/local/executor.go. [1] [2] [3] [4]
New Functionalities:
Introduced OllamaConfig and OllamaConfigUpdate methods to the ILocalModelController interface and implemented them in controller/ai-local/iml.go. These methods allow for getting and updating the Ollama configuration. [1] [2]
Added functionality to automatically subscribe all applications to new services in the Create method of controller/service/iml.go.
Refactoring:
Refactored the initialization of the Ollama client by replacing the static address with a ResetOllamaAddress function in ai-provider/local/local.go.
Removed unused code and imports, such as the newAIUpstream function and upstream_dto import in controller/service/iml.go. [1] [2]
Codebase Simplification:
Simplified the OnInit method in controller/system/iml.go by consolidating the creation of default entities and adding subscription logic. [1] [2] [3]
Additional Changes:
Added new imports and modules to support the new functionalities and refactoring efforts. [1] [2] [3]
* fix: System Settings - General After changing the interface language, the internal pages do not automatically follow the language switch
* fix: login page language error
APIPark 是全球首个开源企业级 API 开放平台，帮助组织快速构建企业内部 API 门户/市场，享受极致的转发性能、API 可观测性、服务治理、多租户管理、订阅审批流程等诸多好处。
<b>🦄 APIPark is an open-source, all-in-one AI gateway and API developer portal, helping developers and enterprises easily manage, integrate, and deploy AI services. APIPark is open-sourced under the Apache 2.0 license, which means it's free for commercial use!</b>
<br>
# ✨ 快速开始
APIPark 致力于为全球企业提供一站式 API 开放与接入产品，打造新一代 API 资产治理标准。APIPark 使用 Apache 2.0 协议开源。
✨ With APIPark, you can:
1. Quickly connect to 100+ AI models, supporting all mainstream AI Companies!
2. Combine AI models and prompt templates into APIs, such as creating a sentiment analysis API, translation API, or data analysis API based on OpenAI GPT-4 and some custom prompts.
3. Standardize the data format of all AI API requests, so switching AI models or modifying prompts won’t affect your APP or microservices, simplifying your AI usage and maintenance costs.
4. Share APIs within the team through APIPark's developer portal.
5. Manage calling applications and API keys to ensure your API's security and stability.
6. Monitor your AI API usage with clear charts.
7. Quickly export API request logs to third-party logging platforms.
APIPark 致力于解决企业在 API 管理中面临的几大关键挑战：
- 复杂的 API 调用关系：简化了复杂系统架构中的 API 交互。
- 数据使用跟踪：提供全面的 API 使用监控和报告。
- 合规管理：确保 API 符合组织和法规标准。
- 故障检测和排查：简化系统问题的识别和解决。
- 量化数据资产价值：提升数据资产的可见性和估值。
<br>
✨ APIPark is also a powerful cloud-native API gateway:
1. It outperforms Nginx with higher performance, supports cluster deployment, and handles large-scale traffic.
2. Share REST APIs within the team, manage API call relationships, and prevent management costs and data breaches caused by chaotic API calls.
<br>
😍APIPark 部署非常简单，仅需一句命令行即可在 5 分钟内部署好你的 API 资产开放平台。
# 💌 Why Did We Build APIPark?
Before building APIPark, we spent seven years developing an API development and automated testing platform with over 1 million developer users, 500+ enterprise customers, and multi-million-dollar investment from Sequoia Capital.
As AI and Agents evolved, we noticed many enterprises wanted to integrate AI into both internal and third-party APIs, enabling AI agents to perform more complex tasks beyond being just knowledge-based Q&A bots. Hence, we built APIPark—your all-in-one AI gateway and API developer portal to accelerate your AI API development and quickly build your product or AI agent!
<br>
# ✨ Quick Start
APIPark is designed to solve the following problems:
- Seamlessly connect to various AI models and package these AI capabilities into APIs for easy calling, significantly reducing the barrier to using AI models.
- Manage complex AI & API call relationships.
- Manage API creation, monitoring, and security.
- Fault detection and troubleshooting: Simplifying system issue identification and resolution.
- Quantify data asset value: Enhance the visibility and valuation of data assets.
<br>
😍 Deploying APIPark is incredibly simple. With just one command line, you can deploy your AI gateway and API developer portal in under 5 minutes.
To achieve this goal, we plan to add new features to APIPark, including:
1. Integrating with API marketplaces such as Postman, RapidAPI, APISpace, APILayer, etc. You can directly use APIs from various API marketplaces through APIPark and make them smarter using AI.
2. Integrating AI Agents such as Langchain, AgentGPT, Auto-GPT, Dify, and more, allowing AI Agents to access your internal or third-party APIs through APIPark to complete more complex tasks.
3. Intelligent API orchestration: APIPark will provide a unified API entry point and automatically orchestrate multiple APIs to fulfill your requests based on the API content.
APIPark uses the Apache 2.0 License. For more details, please refer to the LICENSE file.
<br>
## 💌联系我们
对于企业级功能和专业技术支持，请联系售前专家进行个性化演示、定制方案和获取报价。
# 💌 Contact Us
For enterprise-level features and professional technical support, contact our pre-sales experts for personalized demos, customized solutions, and pricing.
- Website: https://apipark.com
- Email: contact@apipark.com
🙏 A big thanks to everyone who helped shape APIPark. We are thrilled to hear the community’s thoughts! Let’s make the world of APIs and AI stronger and more fun together. 🎉
- 网站: https://apipark.com
- 电子邮件: dev@apipark.com
<br>
感谢您选择 APIPark，下一代 API 资产治理平台。
# 🤝 Partner
- [Cursor](https://www.cursor.com/): Cursor is an AI-powered code editor that integrates artificial intelligence directly into the coding workflow, offering features like intelligent next edit suggestions, deep codebase understanding for relevant answers, and natural language editing to streamline development tasks and boost developer productivity.
- [Dify](https://dify.ai/): Dify is a leading Agentic AI Development Platform that provides a comprehensive suite of tools for building and extending AI applications, offering everything needed for agentic workflows, RAG pipelines, integrations, and observability, while allowing users to amplify their applications with various global Large Language Models (LLMs) and versatile plugins.
- [Trae](https://www.trae.ai/): Trae is an AI-native Integrated Development Environment (IDE) product that aims to embody the concept of “The Real AI Engineer” through intelligent productivity, seamlessly integrating into the development process to enhance quality and efficiency, featuring a chat-based interaction interface and supporting code generation and assistance.
- [Windsurf](https://windsurf.com/): Windsurf is an AI code editor designed to provide a seamless and limitless flow for developers, introducing a new purpose-built IDE that leverages AI to enhance coding with features like "Cascade" for deep codebase understanding, "Windsurf Tab" for intelligent autocompletion, and "Memories" for remembering important aspects of the codebase.
- [Coze](https://www.coze.com/): Coze is a next-generation AI application and chatbot development platform by ByteDance, empowering users to easily create and deploy powerful AI chatbots across various platforms with a no-code bot builder, integrated workflow logic, access to proprietary data, and simplified creation through pre-built plugins, knowledge bases, and workflows.
- [Claude Code](https://www.anthropic.com/claude-code): Claude Code is a command-line AI tool by Anthropic that embeds the Claude Opus 4 model directly into the user’s terminal, providing deep codebase awareness, the ability to edit files and execute commands, and making coordinated changes across multiple files, all while integrating seamlessly with popular IDEs and leveraging existing test suites.
- [Flowith](https://flowith.io/): Flowith is an AI creation workspace designed to revolutionize productivity and deep work by transforming knowledge and streamlining tasks through a multi-thread interface powered by advanced AI agents, offering an intuitive canvas-based user experience unlike traditional chat-based AI tools, and including a 24/7 operational version for complex tasks.
- [OpenManus](https://github.com/FoundationAgents/OpenManus): OpenManus is an open-source framework dedicated to building general AI agents, aiming to provide a platform where users can create and deploy their own agents without an invite code, supporting multi-agent capabilities, and requiring configuration for Large Language Model (LLM) APIs while integrating with browser automation tools.
- [Fellou](https://fellou.ai/): Fellou is an innovative Agentic Browser designed to transcend traditional web browsing by actively performing actions on behalf of the user, automating the entire process of information gathering and insight delivery, and excelling in in-depth research with seamless integrations with popular tools like Notion and LinkedIn.
- [Genspark](https://www.genspark.ai/): Genspark is an ultimate all-in-one AI companion offering a comprehensive suite of tools like AI Slides, AI Sheets, and AI Chat, designed to enhance various aspects of productivity and content creation, with personalized tools and AI Pods for generating content from diverse sources.
- [TEN](https://github.com/TEN-framework/ten-framework): TEN (The Embodied Narrator) is an open-source framework for building real-time, multimodal conversational voice AI agents, including components like TEN Framework, TEN Turn Detection, TEN Agent, TMAN Designer, and TEN Portal, offering features like Real-time Avatar, seamless MCP integration, real-time hardware communication, and vision/screenshare detection.
- [ChatGPT](https://chatgpt.com/): ChatGPT is an AI chatbot developed by OpenAI, built upon large language models like GPT-3.5 and GPT-4, designed to generate human-like conversational dialogue, understand context, answer follow-up questions, and integrate with various platforms for enhanced productivity through advanced language understanding, generation, and multilingual capabilities.
- [LangChain](https://www.langchain.com/): LangChain is a robust platform engineered for the development of reliable agents and Large Language Model (LLM) applications, offering a comprehensive product suite that seamlessly integrates various tools across the entire application development lifecycle, including LangGraph, LangSmith, and the LangGraph Platform, with functionalities for code generation, automation, and AI Search.
- [LEMON AI](https://lemonai.cc/): Lemon AI is the first Full-stack, Open-source, Agentic AI framework, offering a fully local alternative to platforms like Manus & Genspark AI. It features an integrated Code Interpreter VM sandbox for safe execution.
- [LobeChat](https://lobehub.com/): LobeHub offers LobeChat, a personal LLM productivity tool designed to elevate the user experience beyond traditional chatbots by empowering individuals to build personal AI agents and professional teams, supporting a wide array of LLMs, offering a simple chat interface, visual recognition, voice interaction, a rich plugin ecosystem, and knowledge base functionalities.
- [VS Code](https://code.visualstudio.com/): Visual Studio Code (VS Code) is a widely popular, free, and open-source code editor by Microsoft, renowned for its extensibility and customization, supporting vast programming languages, and integrating AI capabilities like intelligent next edit suggestions and an advanced “agent mode” for complex tasks, with broad compatibility with various AI models.
- [XRoute](https://xroute.ai): The Unified Interface For LLMs, provides better prices, better throughput, and no subscription.
- [XPack MCP Marketplace](https://github.com/xpack-ai/XPack-MCP-Marketplace): The world's first open-source MCP monetization platform, transform any OpenAPI into a monetizable MCP server and build your own API marketplace in just 10 minutes. Everything is open-source and ready for commercial use.
- [MemU](https://github.com/NevaMind-AI/memU): MemU is an open-source memory framework for AI companions
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
- name:temperature
use_template:temperature
required:false
type:float
default:1
min:0.0
max:1.0
help:
zh_Hans:生成内容的随机性。
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
- name:temperature
use_template:temperature
required:false
type:float
default:1
min:0.0
max:1.0
help:
zh_Hans:生成内容的随机性。
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
- name:temperature
use_template:temperature
required:false
type:float
default:1
min:0.0
max:1.0
help:
zh_Hans:生成内容的随机性。
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
- name:temperature
use_template:temperature
required:false
type:float
default:1
min:0.0
max:1.0
help:
zh_Hans:生成内容的随机性。
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
- name:temperature
use_template:temperature
required:false
type:float
default:1
min:0.0
max:1.0
help:
zh_Hans:生成内容的随机性。
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
- name:temperature
use_template:temperature
required:false
type:float
default:1
min:0.0
max:1.0
help:
zh_Hans:生成内容的随机性。
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
- name:temperature
use_template:temperature
required:false
type:float
default:1
min:0.0
max:1.0
help:
zh_Hans:生成内容的随机性。
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
- name:temperature
use_template:temperature
required:false
type:float
default:1
min:0.0
max:1.0
help:
zh_Hans:生成内容的随机性。
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
- name:temperature
use_template:temperature
required:false
type:float
default:1
min:0.0
max:1.0
help:
zh_Hans:生成内容的随机性。
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
- name:temperature
use_template:temperature
required:false
type:float
default:1
min:0.0
max:1.0
help:
zh_Hans:生成内容的随机性。
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
- name:temperature
use_template:temperature
required:false
type:float
default:1
min:0.0
max:1.0
help:
zh_Hans:生成内容的随机性。
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
- name:temperature
use_template:temperature
required:false
type:float
default:1
min:0.0
max:1.0
help:
zh_Hans:生成内容的随机性。
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
zh_Hans:停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
en_US:The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
- name:temperature
use_template:temperature
required:false
type:float
default:1
min:0.0
max:1.0
help:
zh_Hans:生成内容的随机性。
en_US:The amount of randomness injected into the response.
- name:top_p
required:false
type:float
default:0.999
min:0.000
max:1.000
help:
zh_Hans:在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
en_US:In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
- name:top_k
required:false
type:int
default:0
min:0
# tip docs from aws has error, max value is 500
max:500
help:
zh_Hans:对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
en_US:Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
- name:response_format
use_template:response_format
pricing:
input:'0.003'
output:'0.015'
unit:'0.001'
currency:USD
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Dot.L
