llms

LLM classes provide access to the large language model (LLM) APIs and services.

Class hierarchy:

BaseLanguageModel --> BaseLLM --> LLM --> <name>  # Examples: AI21, HuggingFaceHub, OpenAI

Main helpers:

LLMResult, PromptValue,
CallbackManagerForLLMRun, AsyncCallbackManagerForLLMRun,
CallbackManager, AsyncCallbackManager,
AIMessage, BaseMessage

Classes

llms.ai21.AI21	AI21 large language models.
llms.ai21.AI21PenaltyData	Parameters for AI21 penalty data.
llms.aleph_alpha.AlephAlpha	Aleph Alpha large language models.
llms.amazon_api_gateway.AmazonAPIGateway	Amazon API Gateway to access LLM models hosted on AWS.
llms.amazon_api_gateway.ContentHandlerAmazonAPIGateway()	Adapter to prepare the inputs from Langchain to a format that LLM model expects.
llms.anyscale.Anyscale	Anyscale large language models.
llms.aphrodite.Aphrodite	Aphrodite language model.
llms.arcee.Arcee	Arcee's Domain Adapted Language Models (DALMs).
llms.aviary.Aviary	Aviary hosted models.
llms.aviary.AviaryBackend(backend_url, bearer)	Aviary backend.
llms.azureml_endpoint.AzureMLBaseEndpoint	Azure ML Online Endpoint models.
llms.azureml_endpoint.AzureMLEndpointApiType(value)	Azure ML endpoints API types.
llms.azureml_endpoint.AzureMLEndpointClient(...)	AzureML Managed Endpoint client.
llms.azureml_endpoint.AzureMLOnlineEndpoint	Azure ML Online Endpoint models.
llms.azureml_endpoint.ContentFormatterBase()	Transform request and response of AzureML endpoint to match with required schema.
llms.azureml_endpoint.CustomOpenAIContentFormatter()	Content formatter for models that use the OpenAI like API scheme.
llms.azureml_endpoint.DollyContentFormatter()	Content handler for the Dolly-v2-12b model
llms.azureml_endpoint.GPT2ContentFormatter()	Content handler for GPT2
llms.azureml_endpoint.HFContentFormatter()	Content handler for LLMs from the HuggingFace catalog.
llms.azureml_endpoint.LlamaContentFormatter()	Deprecated: Kept for backwards compatibility
llms.azureml_endpoint.OSSContentFormatter()	Deprecated: Kept for backwards compatibility
llms.baichuan.BaichuanLLM	Baichuan large language models.
llms.baidu_qianfan_endpoint.QianfanLLMEndpoint	Baidu Qianfan completion model integration.
llms.bananadev.Banana	Banana large language models.
llms.baseten.Baseten	Baseten model
llms.beam.Beam	Beam API for gpt2 large language model.
llms.bedrock.BedrockBase	Base class for Bedrock models.
llms.bedrock.LLMInputOutputAdapter()	Adapter class to prepare the inputs from Langchain to a format that LLM model expects.
llms.bigdl_llm.BigdlLLM	Wrapper around the BigdlLLM model
llms.bittensor.NIBittensorLLM	NIBittensor LLMs
llms.cerebriumai.CerebriumAI	CerebriumAI large language models.
llms.chatglm.ChatGLM	ChatGLM LLM service.
llms.chatglm3.ChatGLM3	ChatGLM3 LLM service.
llms.clarifai.Clarifai	Clarifai large language models.
llms.cloudflare_workersai.CloudflareWorkersAI	Cloudflare Workers AI service.
llms.ctransformers.CTransformers	C Transformers LLM models.
llms.ctranslate2.CTranslate2	CTranslate2 language model.
llms.deepinfra.DeepInfra	DeepInfra models.
llms.deepsparse.DeepSparse	Neural Magic DeepSparse LLM interface.
llms.edenai.EdenAI	EdenAI models.
llms.exllamav2.ExLlamaV2	ExllamaV2 API.
llms.fake.FakeListLLM	Fake LLM for testing purposes.
llms.fake.FakeStreamingListLLM	Fake streaming list LLM for testing purposes.
llms.forefrontai.ForefrontAI	ForefrontAI large language models.
llms.friendli.BaseFriendli	Base class of Friendli.
llms.friendli.Friendli	Friendli LLM.
llms.gigachat.GigaChat	GigaChat large language models API.
llms.gooseai.GooseAI	GooseAI large language models.
llms.gpt4all.GPT4All	GPT4All language models.
llms.gradient_ai.GradientLLM	Gradient.ai LLM Endpoints.
llms.gradient_ai.TrainResult	Train result.
llms.human.HumanInputLLM	User input as the response.
llms.ipex_llm.IpexLLM	IpexLLM model.
llms.javelin_ai_gateway.JavelinAIGateway	Javelin AI Gateway LLMs.
llms.javelin_ai_gateway.Params	Parameters for the Javelin AI Gateway LLM.
llms.koboldai.KoboldApiLLM	Kobold API language model.
llms.konko.Konko	Konko AI models.
llms.layerup_security.LayerupSecurity	Layerup Security LLM service.
llms.llamacpp.LlamaCpp	llama.cpp model.
llms.llamafile.Llamafile	Llamafile lets you distribute and run large language models with a single file.
llms.manifest.ManifestWrapper	HazyResearch's Manifest library.
llms.minimax.Minimax	Minimax large language models.
llms.minimax.MinimaxCommon	Common parameters for Minimax large language models.
llms.mlflow.Mlflow	MLflow LLM service.
llms.mlflow_ai_gateway.MlflowAIGateway	MLflow AI Gateway LLMs.
llms.mlflow_ai_gateway.Params	Parameters for the MLflow AI Gateway LLM.
llms.mlx_pipeline.MLXPipeline	MLX Pipeline API.
llms.modal.Modal	Modal large language models.
llms.moonshot.Moonshot	Moonshot large language models.
llms.moonshot.MoonshotCommon	Common parameters for Moonshot LLMs.
llms.mosaicml.MosaicML	MosaicML LLM service.
llms.nlpcloud.NLPCloud	NLPCloud large language models.
llms.oci_data_science_model_deployment_endpoint.BaseOCIModelDeployment	Base class for LLM deployed on OCI Data Science Model Deployment.
llms.oci_data_science_model_deployment_endpoint.OCIModelDeploymentLLM	LLM deployed on OCI Data Science Model Deployment.
llms.oci_data_science_model_deployment_endpoint.OCIModelDeploymentTGI	OCI Data Science Model Deployment TGI Endpoint.
llms.oci_data_science_model_deployment_endpoint.OCIModelDeploymentVLLM	VLLM deployed on OCI Data Science Model Deployment
llms.oci_data_science_model_deployment_endpoint.ServerError	Raises when encounter server error when making inference.
llms.oci_data_science_model_deployment_endpoint.TokenExpiredError	Raises when token expired.
llms.oci_generative_ai.CohereProvider()
llms.oci_generative_ai.MetaProvider()
llms.oci_generative_ai.OCIAuthType(value[, ...])	OCI authentication types as enumerator.
llms.oci_generative_ai.OCIGenAI	OCI large language models.
llms.oci_generative_ai.OCIGenAIBase	Base class for OCI GenAI models
llms.oci_generative_ai.Provider()
llms.octoai_endpoint.OctoAIEndpoint	OctoAI LLM Endpoints - OpenAI compatible.
llms.ollama.OllamaEndpointNotFoundError	Raised when the Ollama endpoint is not found.
llms.opaqueprompts.OpaquePrompts	LLM that uses OpaquePrompts to sanitize prompts.
llms.openai.BaseOpenAI	Base OpenAI large language model class.
llms.openllm.OpenLLM	OpenAI's compatible API client for OpenLLM server
llms.openlm.OpenLM	OpenLM models.
llms.outlines.Outlines	LLM wrapper for the Outlines library.
llms.pai_eas_endpoint.PaiEasEndpoint	Langchain LLM class to help to access eass llm service.
llms.petals.Petals	Petals Bloom models.
llms.pipelineai.PipelineAI	PipelineAI large language models.
llms.predibase.Predibase	Use your Predibase models with Langchain.
llms.promptlayer_openai.PromptLayerOpenAI	PromptLayer OpenAI large language models.
llms.promptlayer_openai.PromptLayerOpenAIChat	PromptLayer OpenAI large language models.
llms.replicate.Replicate	Replicate models.
llms.rwkv.RWKV	RWKV language models.
llms.sagemaker_endpoint.ContentHandlerBase()	Handler class to transform input from LLM to a format that SageMaker endpoint expects.
llms.sagemaker_endpoint.LLMContentHandler()	Content handler for LLM class.
llms.sagemaker_endpoint.LineIterator(stream)	Parse the byte stream input.
llms.sagemaker_endpoint.SagemakerEndpoint	Sagemaker Inference Endpoint models.
llms.sambanova.SambaNovaCloud	SambaNova Cloud large language models.
llms.sambanova.SambaStudio	SambaStudio large language models.
llms.self_hosted.SelfHostedPipeline	Model inference on self-hosted remote hardware.
llms.self_hosted_hugging_face.SelfHostedHuggingFaceLLM	HuggingFace Pipeline API to run on self-hosted remote hardware.
llms.solar.Solar	Solar large language models.
llms.solar.SolarCommon	Common configuration for Solar LLMs.
llms.sparkllm.SparkLLM	iFlyTek Spark completion model integration.
llms.stochasticai.StochasticAI	StochasticAI large language models.
llms.symblai_nebula.Nebula	Nebula Service models.
llms.textgen.TextGen	Text generation models from WebUI.
llms.titan_takeoff.Device(value[, names, ...])	The device to use for inference, cuda or cpu
llms.titan_takeoff.ReaderConfig	Configuration for the reader to be deployed in Titan Takeoff API.
llms.titan_takeoff.TitanTakeoff	Titan Takeoff API LLMs.
llms.tongyi.Tongyi	Tongyi completion model integration.
llms.vllm.VLLM	VLLM language model.
llms.vllm.VLLMOpenAI	vLLM OpenAI-compatible API client
llms.volcengine_maas.VolcEngineMaasBase	Base class for VolcEngineMaas models.
llms.volcengine_maas.VolcEngineMaasLLM	volc engine maas hosts a plethora of models.
llms.weight_only_quantization.WeightOnlyQuantPipeline	Weight only quantized model.
llms.writer.Writer	Writer large language models.
llms.xinference.Xinference	Xinference large-scale model inference service.
llms.yandex.YandexGPT	Yandex large language models.
llms.yi.YiLLM	Yi large language models.
llms.you.You	Wrapper around You.com's conversational Smart and Research APIs.
llms.yuan2.Yuan2	Yuan2.0 language models.

Functions

llms.anyscale.create_llm_result(choices, ...)	Create the LLMResult from the choices and prompts.
llms.anyscale.update_token_usage(keys, ...)	Update token usage.
llms.aviary.get_completions(model, prompt[, ...])	Get completions from Aviary models.
llms.aviary.get_models()	List available models
llms.cohere.acompletion_with_retry(llm, **kwargs)	Use tenacity to retry the completion call.
llms.cohere.completion_with_retry(llm, **kwargs)	Use tenacity to retry the completion call.
llms.databricks.get_default_api_token()	Get the default Databricks personal access token.
llms.databricks.get_default_host()	Get the default Databricks workspace hostname.
llms.databricks.get_repl_context()	Get the notebook REPL context if running inside a Databricks notebook.
llms.fireworks.acompletion_with_retry(llm, ...)	Use tenacity to retry the completion call.
llms.fireworks.acompletion_with_retry_batching(...)	Use tenacity to retry the completion call.
llms.fireworks.acompletion_with_retry_streaming(...)	Use tenacity to retry the completion call for streaming.
llms.fireworks.completion_with_retry(llm, ...)	Use tenacity to retry the completion call.
llms.fireworks.completion_with_retry_batching(...)	Use tenacity to retry the completion call.
llms.fireworks.conditional_decorator(...)	Conditionally apply a decorator.
llms.google_palm.completion_with_retry(llm, ...)	Use tenacity to retry the completion call.
llms.koboldai.clean_url(url)	Remove trailing slash and /api from url if present.
llms.layerup_security.default_guardrail_violation_handler(...)	Default guardrail violation handler.
llms.loading.load_llm(file, **kwargs)	Load LLM from a file.
llms.loading.load_llm_from_config(config, ...)	Load LLM from Config Dict.
llms.openai.acompletion_with_retry(llm[, ...])	Use tenacity to retry the async completion call.
llms.openai.completion_with_retry(llm[, ...])	Use tenacity to retry the completion call.
llms.openai.update_token_usage(keys, ...)	Update token usage.
llms.symblai_nebula.completion_with_retry(...)	Use tenacity to retry the completion call.
llms.symblai_nebula.make_request(self, prompt)	Generate text from the model.
llms.tongyi.agenerate_with_last_element_mark(...)	Generate elements from an async iterable, and a boolean indicating if it is the last element.
llms.tongyi.astream_generate_with_retry(llm, ...)	Async version of stream_generate_with_retry.
llms.tongyi.check_response(resp)	Check the response from the completion call.
llms.tongyi.generate_with_last_element_mark(...)	Generate elements from an iterable, and a boolean indicating if it is the last element.
llms.tongyi.generate_with_retry(llm, **kwargs)	Use tenacity to retry the completion call.
llms.tongyi.stream_generate_with_retry(llm, ...)	Use tenacity to retry the completion call.
llms.utils.enforce_stop_tokens(text, stop)	Cut off the text as soon as any stop words occur.
llms.vertexai.acompletion_with_retry(llm, prompt)	Use tenacity to retry the completion call.
llms.vertexai.completion_with_retry(llm, prompt)	Use tenacity to retry the completion call.
llms.vertexai.is_codey_model(model_name)	Return True if the model name is a Codey model.
llms.vertexai.is_gemini_model(model_name)	Return True if the model name is a Gemini model.
llms.yandex.acompletion_with_retry(llm, **kwargs)	Use tenacity to retry the async completion call.
llms.yandex.completion_with_retry(llm, **kwargs)	Use tenacity to retry the completion call.

Deprecated classes

llms.anthropic.Anthropic
llms.bedrock.Bedrock
llms.cohere.BaseCohere
llms.cohere.Cohere
llms.databricks.Databricks
llms.fireworks.Fireworks
llms.google_palm.GooglePalm
llms.huggingface_endpoint.HuggingFaceEndpoint
llms.huggingface_hub.HuggingFaceHub
llms.huggingface_pipeline.HuggingFacePipeline
llms.huggingface_text_gen_inference.HuggingFaceTextGenInference
llms.ollama.Ollama
llms.openai.AzureOpenAI
llms.openai.OpenAI
llms.openai.OpenAIChat
llms.predictionguard.PredictionGuard
llms.together.Together
llms.vertexai.VertexAI
llms.vertexai.VertexAIModelGarden
llms.watsonxllm.WatsonxLLM