Model Garden

The model garden is a repository of pre-trained and fine-tuned large language models (LLMs) that can be used by developers to build applications. It provides a convenient way for developers to access the power of LLMs without training their own models from scratch.

Orquesta's model garden typically includes various LLMs, each with its own strengths and weaknesses. This allows developers to choose the right LLM for their specific needs. For example, a developer building a chatbot might choose an LLM that is good at generating conversational language. A developer building a translation tool might choose an LLM that is good at translating between different languages.

Supported LLM Providers

Orquesta supports a long list of LLM providers and models to provide developers with the best service for prompt engineering and LLM Ops.

Orquesta is LLM-agnostic. Therefore, we want to get out of your way in working with public, private and custom LLM Providers and Models. Within the Model Garden, you can enable providers and specific models your product works with.

Models by Provider

The following are the models available for each LLM Provider, and all these models can be used for prompt engineering and prompt management.

Anthropic:

claude-1, claude-1-100k, claude-instant-1-100k, claude-instant-1, claude-2

Azure:

gpt-35-turbo, gpt-35-turbo-16k, gpt-35-turbo-instruct, gpt-4, gpt-4-32k

Cohere:

command, command-light, command-light-nightly, command-nightly, coral-command, 
coral-command-light, coral-command-light-nightly, coral-command-nightly

Google:

chat-bison@001, codechat-bison@001, code-bison@001, code-gecko@001, 
text-bison

Huggingface:

EleutherAI/gpt-neox-20b, bigscience/bloomz, bigcode/santacoder, 
bigscience/bloom, google/flan-t5-xxl, google/flan-ul2, 
OpenAssistant/oasst-sft-4-pythia-12b-epoch-3.5

OpenAI:

gpt-3.5-turbo, gpt-3.5-turbo-0301, gpt-3.5-turbo-0613, gpt-3.5-turbo-16k, 
gpt-3.5-turbo-16k-0613, gpt-3.5-turbo-instruct, gpt-4, gpt-4-0314, 
gpt-4-0613, gpt-4-32k, gpt-4-32k-0613

Replicate:

llama-7b, stablelm-tuned-alpha-7b, vicuna-13b

For a more detailed overview of all the supported LLM Providers and models available in Orquesta's model garden, you can check it out here.

Activating a Model

The steps in activating a model are straightforward. Click on Model Garden; you will see a list of all the available LLM Providers and models, and you can toggle the models your teams can use across your workspace.

Gateway support

The gateway is an abstraction layer used to integrate different LLMs using Orquesta. You can find all the models supported in the gateway (it has a purple icon) in the model garden, when you select these models you will be able to create prompts and endpoints that go straight to the model without an API call.  

The models supported in the Gateway can be used as a primary model and a fallback model in the Endpoint. The remaining Providers and models that are not supported in the Gateway yet such as Claude, Google, Replicate, some Cohere models (base and base-light), and some OpenAI models (text-davinci-001, text-davinci-002, text-davinci-003) can be used in the Prompt for prompt management.

Benefits of the Model Garden

• Convenience: The model gardens provide a convenient way for developers to access the power of LLMs without having to train their own models from scratch.
  
  

• Choice: It typically includes various LLMs, each with its own strengths and weaknesses, allowing developers to choose the right LLM for their specific needs.