jyapayne/langflow
1
0
Fork 0
Code Issues Pull requests Projects Releases 1 Packages Wiki Activity Actions
langflow/docs/docs/components/models.mdx
Carlos Coelho 2587849fea
 Update docs (#1567) 
* Add new documentation files and update package dependencies

* Refactor tweak application logic in process_tweaks function

* Add dynamic function creation and execution helpers

* Refactor build method to be asynchronous

* Add FlowToolComponent to handle flows as tools

* Update RunFlowComponent to include a method for updating build config

* Fix duplicated first layer results

* Refactor vertex building and streaming endpoints

* Add base_name attribute to Vertex class

* Refactor flow.py to generate dynamic flow functions and build schemas

* Refactor FlowToolComponent in FlowTool.py

* Add JSONInputComponent to load JSON object as input

* Update render_tool_description method in XMLAgent.py

* Refactor XMLAgentComponent.render_tool_description() method

* Refactor SearchApi.py to include typing and handle empty records

* Refactor SearchApi class to simplify code

* Add SearchApi and SearchApiTool components

* Refactor ServiceFactory and Dependencies (#1560)

* Update dependencies for OpenTelemetry

* Update service dependency logic and add first version of telemetry service

* Remove telemetry service and related code

* Update cache service references

* Refactor imports in env.py

* Refactor code for initializing services and socketio server

* Refactor parameterComponent to use inline button_text

* Refactor build_vertex method and add RunnableVerticesManager class

* Add import statement and update build_vertex function

* Add import statement for SettingsService in MonitorServiceFactory.create() method

* Refactor build_schema_from_inputs to use display_name and description for field names and descriptions respectively

* Refactor graph building and running logic

* Update input type mappings and function arguments

* Update default values for input types in flow.py

* Remove console.log statement in flowStore.ts

* Add vertices_to_run field to VerticesOrderResponse

* Add input_value parameter to chain components

* Refactor CSVAgent build method to include handle_parse_errors parameter

* Add agent_type parameter to CSVAgent build method

* Update model imports in component files

* Add LCAgentComponent and XMLAgentComponent

* Add "agents" category to NATIVE_CATEGORIES

* Refactor model.py to support chat models

* Add system_message parameter to model components

* Update CSVAgent.py: handle_parsing_errors and agent_type options

* Add ping animation to update button

* Fix encryption and decryption of API keys

* Update CSVAgentComponent constructor

* Refactor inputs parameter to inputs_dict in build_vertex function

* Removes "component" table and drops "flowstyle" table

* Delete component model and init files

* Removes "flowstyle" table and drops "user" table index

* Add typing import to CohereModel.py

* Fix ShareModal rendering issue

* Update models docs

* Changed vector-stores docs

* Update component documentation

* Add AstraDB and AstraDBSearch components for AstraDB Vector Store docs

* Rename GetNotified to Listen

* Update GetNotifiedComponent import

* Remove unused imports in flow-runner.mdx and features.mdx

* Add new documentation files and update existing files

* Update package versions in package-lock.json

* Remove unused files

* Delete run-flow.mdx file

* Update topics

* Add new file run-flow.mdx

---------

Co-authored-by: Gabriel Luiz Freitas Almeida <gabriel@logspace.ai>
Co-authored-by: anovazzi1 <otavio2204@gmail.com>
2024-03-26 13:55:54 -03:00

464 lines
No EOL
17 KiB
Text
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

import Admonition from '@theme/Admonition';
# Models
<Admonition type="caution" icon="🚧" title="ZONE UNDER CONSTRUCTION">
<p>
We appreciate your understanding as we polish our documentation – it may contain some rough edges. Share your feedback or report issues to help us improve! 🛠️📝
</p>
</Admonition>
### AmazonBedrock
This component facilitates the generation of text using the LLM (Large Language Model) model from Amazon Bedrock.
**Params**
- **Input Value:** Specifies the input text for text generation.
- **System Message (Optional):** A system message to pass to the model.
- **Model ID (Optional):** Specifies the model ID to be used for text generation. Defaults to _`"anthropic.claude-instant-v1"`_. Available options include:
- _`"ai21.j2-grande-instruct"`_
- _`"ai21.j2-jumbo-instruct"`_
- _`"ai21.j2-mid"`_
- _`"ai21.j2-mid-v1"`_
- _`"ai21.j2-ultra"`_
- _`"ai21.j2-ultra-v1"`_
- _`"anthropic.claude-instant-v1"`_
- _`"anthropic.claude-v1"`_
- _`"anthropic.claude-v2"`_
- _`"cohere.command-text-v14"`_
- **Credentials Profile Name (Optional):** Specifies the name of the credentials profile.
- **Region Name (Optional):** Specifies the region name.
- **Model Kwargs (Optional):** Additional keyword arguments for the model.
- **Endpoint URL (Optional):** Specifies the endpoint URL.
- **Streaming (Optional):** Specifies whether to stream the response from the model. Defaults to _`False`_.
- **Cache (Optional):** Specifies whether to cache the response.
- **Stream (Optional):** Specifies whether to stream the response from the model. Defaults to _`False`_.
<Admonition type="note" title="Note">
<p>
Ensure that necessary credentials are provided to connect to the Amazon Bedrock API. If connection fails, a ValueError will be raised.
</p>
</Admonition>
---
### AnthropicLLM
This component allows the generation of text using Anthropic Chat&Completion large language models.
**Params**
- **Model Name:** Specifies the name of the Anthropic model to be used for text generation. Available options include:
- _`"claude-2.1"`_
- _`"claude-2.0"`_
- _`"claude-instant-1.2"`_
- _`"claude-instant-1"`_
- **Anthropic API Key:** Your Anthropic API key.
- **Max Tokens (Optional):** Specifies the maximum number of tokens to generate. Defaults to _`256`_.
- **Temperature (Optional):** Specifies the sampling temperature. Defaults to _`0.7`_.
- **API Endpoint (Optional):** Specifies the endpoint of the Anthropic API. Defaults to _`"https://api.anthropic.com"`_ if not specified.
- **Input Value:** Specifies the input text for text generation.
- **Stream (Optional):** Specifies whether to stream the response from the model. Defaults to _`False`_.
- **System Message (Optional):** A system message to pass to the model.
For detailed documentation and integration guides, please refer to the [Anthropic Component Documentation](https://python.langchain.com/docs/integrations/chat/anthropic).
---
### AzureChatOpenAI
This component allows the generation of text using the LLM (Large Language Model) model from Azure OpenAI.
**Params**
- **Model Name:** Specifies the name of the Azure OpenAI model to be used for text generation. Available options include:
- _`"gpt-35-turbo"`_
- _`"gpt-35-turbo-16k"`_
- _`"gpt-35-turbo-instruct"`_
- _`"gpt-4"`_
- _`"gpt-4-32k"`_
- _`"gpt-4-vision"`_
- **Azure Endpoint:** Your Azure endpoint, including the resource. Example: `https://example-resource.azure.openai.com/`.
- **Deployment Name:** Specifies the name of the deployment.
- **API Version:** Specifies the version of the Azure OpenAI API to be used. Available options include:
- _`"2023-03-15-preview"`_
- _`"2023-05-15"`_
- _`"2023-06-01-preview"`_
- _`"2023-07-01-preview"`_
- _`"2023-08-01-preview"`_
- _`"2023-09-01-preview"`_
- _`"2023-12-01-preview"`_
- **API Key:** Your Azure OpenAI API key.
- **Temperature (Optional):** Specifies the sampling temperature. Defaults to _`0.7`_.
- **Max Tokens (Optional):** Specifies the maximum number of tokens to generate. Defaults to _`1000`_.
- **Input Value:** Specifies the input text for text generation.
- **Stream (Optional):** Specifies whether to stream the response from the model. Defaults to _`False`_.
- **System Message (Optional):** A system message to pass to the model.
For detailed documentation and integration guides, please refer to the [Azure OpenAI Component Documentation](https://python.langchain.com/docs/integrations/llms/azure_openai).
---
### QianfanChatEndpoint
This component facilitates the generation of text using Baidu Qianfan chat models.
**Params**
- **Model Name:** Specifies the name of the Qianfan chat model to be used for text generation. Available options include:
- _`"ERNIE-Bot"`_
- _`"ERNIE-Bot-turbo"`_
- _`"BLOOMZ-7B"`_
- _`"Llama-2-7b-chat"`_
- _`"Llama-2-13b-chat"`_
- _`"Llama-2-70b-chat"`_
- _`"Qianfan-BLOOMZ-7B-compressed"`_
- _`"Qianfan-Chinese-Llama-2-7B"`_
- _`"ChatGLM2-6B-32K"`_
- _`"AquilaChat-7B"`_
- **Qianfan Ak:** Your Baidu Qianfan access key, obtainable from [here](https://cloud.baidu.com/product/wenxinworkshop).
- **Qianfan Sk:** Your Baidu Qianfan secret key, obtainable from [here](https://cloud.baidu.com/product/wenxinworkshop).
- **Top p (Optional):** Model parameter. Specifies the top-p value. Only supported in ERNIE-Bot and ERNIE-Bot-turbo models. Defaults to _`0.8`_.
- **Temperature (Optional):** Model parameter. Specifies the sampling temperature. Only supported in ERNIE-Bot and ERNIE-Bot-turbo models. Defaults to _`0.95`_.
- **Penalty Score (Optional):** Model parameter. Specifies the penalty score. Only supported in ERNIE-Bot and ERNIE-Bot-turbo models. Defaults to _`1.0`_.
- **Endpoint (Optional):** Endpoint of the Qianfan LLM, required if custom model is used.
- **Input Value:** Specifies the input text for text generation.
- **Stream (Optional):** Specifies whether to stream the response from the model. Defaults to _`False`_.
- **System Message (Optional):** A system message to pass to the model.
---
### Cohere
This component enables text generation using Cohere large language models.
**Params**
- **Cohere API Key:** Your Cohere API key.
- **Max Tokens (Optional):** Specifies the maximum number of tokens to generate. Defaults to _`256`_.
- **Temperature (Optional):** Specifies the sampling temperature. Defaults to _`0.75`_.
- **Input Value:** Specifies the input text for text generation.
- **Stream (Optional):** Specifies whether to stream the response from the model. Defaults to _`False`_.
- **System Message (Optional):** A system message to pass to the model.
---
### CTransformers
This component allows the generation of text using CTransformers large language models.
**Params**
- **Model:** Specifies the CTransformers model to be used for text generation.
- **Model File (Optional):** Path to the model file if using a custom model. Should be a _.bin_ file.
- **Model Type:** Specifies the type of the CTransformers model.
- **Config (Optional):** Additional configuration parameters for the model. It should be provided as a JSON object.
Defaults to:
`{"top_k":40,"top_p":0.95,"temperature":0.8,"repetition_penalty":1.1,"last_n_tokens":64,"seed":-1,"max_new_tokens":256,"stop":"","stream":"False","reset":"True","batch_size":8,"threads":-1,"context_length":-1,"gpu_layers":0}`.
- **Input Value:** Specifies the input text for text generation.
- **Stream (Optional):** Specifies whether to stream the response from the model. Defaults to _`False`_.
- **System Message (Optional):** A system message to pass to the model.
---
### Google Generative AI
This component enables text generation using Google Generative AI.
**Params**
- **Google API Key:** Your Google API key to use for the Google Generative AI.
- **Model:** The name of the model to use. Supported examples are _`"gemini-pro"`_ and _`"gemini-pro-vision"`_.
- **Max Output Tokens (Optional):** The maximum number of tokens to generate.
- **Temperature:** Run inference with this temperature. Must be in the closed interval [0.0, 1.0].
- **Top K (Optional):** Decode using top-k sampling: consider the set of top_k most probable tokens. Must be positive.
- **Top P (Optional):** The maximum cumulative probability of tokens to consider when sampling.
- **N (Optional):** Number of chat completions to generate for each prompt. Note that the API may not return the full n completions if duplicates are generated.
- **Input Value:** The input to the model.
- **Stream (Optional):** Specifies whether to stream the response from the model. Defaults to _`False`_.
- **System Message (Optional):** A system message to pass to the model.
---
### Hugging Face API
This component facilitates text generation using LLM models from the Hugging Face Inference API.
**Params**
- **Endpoint URL:** The URL of the Hugging Face Inference API endpoint. Should be provided along with necessary authentication credentials.
- **Task:** Specifies the task for text generation. Options include _`"text2text-generation"`_, _`"text-generation"`_, and _`"summarization"`_.
- **API Token:** The API token required for authentication with the Hugging Face Hub.
- **Model Keyword Arguments (Optional):** Additional keyword arguments for the model. Should be provided as a Python dictionary.
- **Input Value:** The input text for text generation.
- **Stream (Optional):** Specifies whether to stream the response from the model. Defaults to _`False`_.
- **System Message (Optional):** A system message to pass to the model.
---
### LlamaCpp
The `LlamaCpp` is a component for generating text using the llama.cpp model.
**Params**
- **Model Path:** The path to the llama.cpp model file. This should be provided as a file type input.
- **Input Value:** The input text for text generation.
- **Grammar (Optional):** The grammar for text generation.
- **Cache (Optional):** Specifies whether to cache the generated text.
- **Client (Optional):** The client to use for text generation.
- **Echo (Optional):** Specifies whether to echo the generated text. Defaults to _`False`_.
- **F16 KV:** Specifies whether to use F16 key-value pairs. Defaults to _`True`_.
- **Grammar Path (Optional):** The path to the grammar file.
- **Last N Tokens Size (Optional):** The size of the last N tokens. Defaults to _`64`_.
- **Logits All:** Specifies whether to include logits for all tokens. Defaults to _`False`_.
- **Logprobs (Optional):** The log probabilities for text generation.
- **Lora Base (Optional):** The base URL for Lora.
- **Lora Path (Optional):** The path for Lora.
- **Max Tokens (Optional):** The maximum number of tokens to generate. Defaults to _`256`_.
- **Metadata (Optional):** Additional metadata for the model.
- **Model Kwargs:** Additional keyword arguments for the model. Should be provided as a Python dictionary.
- **N Batch (Optional):** The batch size. Defaults to _`8`_.
- **N Ctx:** The context size. Defaults to _`512`_.
- **N GPU Layers (Optional):** The number of GPU layers.
- **N Parts:** The number of parts.
- **N Threads (Optional):** The number of threads. Defaults to _`1`_.
- **Repeat Penalty (Optional):** The repeat penalty for text generation. Defaults to _`1.1`_.
- **Rope Freq Base:** The base frequency for rope.
- **Rope Freq Scale:** The scale frequency for rope.
- **Seed:** The seed for random generation.
- **Stop (Optional):** The stop words for text generation.
- **Streaming:** Specifies whether to stream the response from the model. Defaults to _`True`_.
- **Suffix (Optional):** The suffix for text generation.
- **Tags (Optional):** The tags for text generation.
- **Temperature (Optional):** The temperature for text generation. Defaults to _`0.8`_.
- **Top K (Optional):** The top K tokens to consider for text generation. Defaults to _`40`_.
- **Top P (Optional):** The top P probability threshold for text generation. Defaults to _`0.95`_.
- **Use Mlock:** Specifies whether to use Mlock. Defaults to _`False`_.
- **Use Mmap (Optional):** Specifies whether to use Mmap. Defaults to _`True`_.
- **Verbose:** Specifies whether to enable verbose mode. Defaults to _`True`_.
- **Vocab Only:** Specifies whether to include vocabulary only.
- **Stream (Optional):** Specifies whether to stream the response from the model. Defaults to _`False`_.
For more information, please refer to the [documentation](https://python.langchain.com/docs/modules/model_io/models/llms/integrations/llamacpp).
---
### ChatOllama
This component facilitates text generation using the Local LLM model for chat with Ollama.
**Params**
- **Base URL:** The endpoint of the Ollama API. Defaults to 'http://localhost:11434' if not specified.
- **Model Name:** The name of the model to use. Refer to [https://ollama.ai/library](https://ollama.ai/library) for more models.
- **Input Value:** The input text for text generation.
- **Mirostat:** Enable/disable Mirostat sampling for controlling perplexity.
- **Mirostat Eta (Optional):** The learning rate for the Mirostat algorithm. (Default: 0.1)
- **Mirostat Tau (Optional):** Controls the balance between coherence and diversity of the output. (Default: 5.0)
- **Repeat Last N (Optional):** How far back the model looks to prevent repetition. (Default: 64, 0 = disabled, -1 = num_ctx)
- **Verbose (Optional):** Whether to print out response text.
- **Cache (Optional):** Enable or disable caching. Defaults to _`False`_.
- **Context Window Size (Optional):** Size of the context window for generating tokens. (Default: 2048)
- **Number of GPUs (Optional):** Number of GPUs to use for computation. (Default: 1 on macOS, 0 to disable)
- **Format (Optional):** Specify the format of the output (e.g., json).
- **Metadata (Optional):** Metadata to add to the run trace.
- **Number of Threads (Optional):** Number of threads to use during computation. (Default: detected for optimal performance)
- **Repeat Penalty (Optional):** Penalty for repetitions in generated text. (Default: 1.1)
- **Stop Tokens (Optional):** List of tokens to signal the model to stop generating text.
- **System (Optional):** System to use for generating text.
- **Tags (Optional):** Tags to add to the run trace.
- **Temperature (Optional):** Controls the creativity of model responses. Defaults to _`0.8`_.
- **Template (Optional):** Template to use for generating text.
- **TFS Z (Optional):** Tail free sampling value. (Default: 1)
- **Timeout (Optional):** Timeout for the request stream.
- **Top K (Optional):** Limits token selection to top K. (Default: 40)
- **Top P (Optional):** Works together with top-k. (Default: 0.9)
- **Stream (Optional):** Specifies whether to stream the response from the model. Defaults to _`False`_.
- **System Message (Optional):** System message to pass to the model.
---
### OpenAIModel
This component facilitates text generation using OpenAI's models.
**Params**
- **Input Value:** The input text for text generation.
- **Max Tokens (Optional):** The maximum number of tokens to generate. Defaults to _`256`_.
- **Model Kwargs (Optional):** Additional keyword arguments for the model. Should be provided as a nested dictionary.
- **Model Name (Optional):** The name of the model to use. Defaults to _`gpt-4-1106-preview`_. Supported options include: _`gpt-4-turbo-preview`_, _`gpt-4-0125-preview`_, _`gpt-4-1106-preview`_, _`gpt-4-vision-preview`_, _`gpt-3.5-turbo-0125`_, _`gpt-3.5-turbo-1106`_.
- **OpenAI API Base (Optional):** The base URL of the OpenAI API. Defaults to _`https://api.openai.com/v1`_.
- **OpenAI API Key (Optional):** The API key for accessing the OpenAI API.
- **Temperature:** Controls the creativity of model responses. Defaults to _`0.7`_.
- **Stream (Optional):** Specifies whether to stream the response from the model. Defaults to _`False`_.
- **System Message (Optional):** System message to pass to the model.
---
### ChatVertexAI
The `ChatVertexAI` is a component for generating text using Vertex AI Chat large language models API.
**Params**
- **Input Value:** The input text for text generation.
- **Credentials:** The JSON file containing the credentials for accessing the Vertex AI Chat API.
- **Project:** The name of the project associated with the Vertex AI Chat API.
- **Examples (Optional):** List of examples to provide context for text generation.
- **Location:** The location of the Vertex AI Chat API service. Defaults to _`us-central1`_.
- **Max Output Tokens:** The maximum number of tokens to generate. Defaults to _`128`_.
- **Model Name:** The name of the model to use. Defaults to _`chat-bison`_.
- **Temperature:** Controls the creativity of model responses. Defaults to _`0.0`_.
- **Top K:** Limits token selection to top K. Defaults to _`40`_.
- **Top P:** Works together with top-k. Defaults to _`0.95`_.
- **Verbose:** Whether to print out response text. Defaults to _`False`_.
- **Stream (Optional):** Specifies whether to stream the response from the model. Defaults to _`False`_.
- **System Message (Optional):** System message to pass to the model.
Reference in a new issue View git blame Copy permalink