|
At the moment, we’re introducing Amazon Bedrock Market, a brand new functionality that offers you entry to over 100 standard, rising, and specialised basis fashions (FMs) via Amazon Bedrock. With this launch, now you can uncover, take a look at, and deploy new fashions from enterprise suppliers comparable to IBM and Nvidia, specialised fashions comparable to Upstages’ Photo voltaic Professional for Korean language processing, and Evolutionary Scale’s ESM3 for protein analysis, alongside Amazon Bedrock general-purpose FMs from suppliers comparable to Anthropic and Meta.
Fashions deployed with Amazon Bedrock Market may be accessed via the identical commonplace APIs because the serverless fashions and, for fashions that are suitable with Converse API, be used with instruments comparable to Amazon Bedrock Brokers and Amazon Bedrock Data Bases.
As generative AI continues to reshape how organizations work, the necessity for specialised fashions optimized for particular domains, languages, or duties is rising. Nonetheless, discovering and evaluating these fashions may be difficult and expensive. You might want to uncover them throughout completely different providers, construct abstractions to make use of them in your purposes, and create complicated safety and governance layers. Amazon Bedrock Market addresses these challenges by offering a single interface to entry each specialised and general-purpose FMs.
Utilizing Amazon Bedrock Market
To get began, within the Amazon Bedrock console, I select Mannequin catalog within the Basis fashions part of the navigation pane. Right here, I can seek for fashions that assist me with a selected use case or language. The outcomes of the search embody each serverless fashions and fashions obtainable in Amazon Bedrock Market. I can filter outcomes by supplier, modality (comparable to textual content, picture, or audio), or process (comparable to classification or textual content summarization).
Within the catalog, there are fashions from organizations like Arcee AI, which builds context-adapted small language fashions (SLMs), and Widn.AI, which supplies multilingual fashions.
For instance, I’m within the IBM Granite fashions and seek for fashions from IBM Knowledge and AI.
I choose Granite 3.0 2B Instruct, a language mannequin designed for enterprise purposes. Selecting the mannequin opens the mannequin element web page the place I can see extra data from the mannequin supplier comparable to highlights in regards to the mannequin, pricing, and utilization together with pattern API calls.
This particular mannequin requires a subscription, and I select View subscription choices.
From the subscription dialog, I assessment pricing and authorized notes. In Pricing particulars, I see the software program value set by the supplier. For this mannequin, there aren’t any further prices on prime of the deployed infrastructure. The Amazon SageMaker infrastructure price is charged individually and may be seen in Amazon SageMaker pricing.
To proceed with this mannequin, I select Subscribe.
After the subscription has been accomplished, which normally takes a couple of minutes, I can deploy the mannequin. For Deployment particulars, I exploit the default settings and the beneficial occasion kind.
I increase the non-obligatory Superior settings. Right here, I can select to deploy in a digital personal cloud (VPC) or specify the AWS Identification and Entry Administration (IAM) service function utilized by the deployment. Amazon Bedrock Market mechanically creates a service function to entry Amazon Easy Storage Service (Amazon S3) buckets the place the mannequin weights are saved, however I can select to make use of an current function.
I hold the default values and full the deployment.
After a couple of minutes, the deployment is In Service and may be reviewed within the Market deployments web page from the navigation pane.
There, I can select an endpoint to view particulars and edit the configuration such because the variety of cases. To check the deployment, I select Open in playground and ask for some poetry.
I also can choose the mannequin from the Chat/textual content web page of the Playground utilizing the brand new Market class the place the deployed endpoints are listed.
In an analogous approach, I can use the mannequin with different instruments comparable to Amazon Bedrock Brokers, Amazon Bedrock Data Bases, Amazon Bedrock Immediate Administration, Amazon Bedrock Guardrails, and mannequin evaluations, by selecting Choose Mannequin and deciding on the Market mannequin endpoint.
The mannequin I used right here is text-to-text, however I can use Amazon Bedrock Market to deploy fashions with completely different modalities. For instance, after I deploy Stability AI Steady Diffusion 3.5 Giant, I can run a fast take a look at within the Amazon Bedrock Picture playground.
The fashions I deployed at the moment are obtainable via the Amazon Bedrock InvokeModel API. When a mannequin is deployed, I can use it with the AWS Command Line Interface (AWS CLI) and any AWS SDKs utilizing the endpoint Amazon Useful resource Identify (ARN) as mannequin ID.
For chat-tuned text-to-text fashions, I also can use the Amazon Bedrock Converse API, which abstracts mannequin variations and allows mannequin switching with a single parameter change.
Issues to know
Amazon Bedrock Market is obtainable within the following AWS Areas: US East (N. Virginia), US East (Ohio), US West (Oregon), Asia Pacific (Mumbai), Asia Pacific (Seoul), Asia Pacific (Singapore), Asia Pacific (Sydney), Asia Pacific (Tokyo), Canada (Central), Europe (Frankfurt), Europe (Eire), Europe (London), Europe (Paris), and South America (São Paulo).
With Amazon Bedrock Market, you pay a software program payment to the third-party mannequin supplier (which may be zero, as within the earlier instance) and a internet hosting payment based mostly on the sort and variety of cases you select on your mannequin endpoints.
Begin shopping the brand new fashions utilizing the Mannequin catalog within the Amazon Bedrock console, go to the Amazon Bedrock Market documentation, and ship suggestions to AWS re:Publish for Amazon Bedrock. You’ll find deep-dive technical content material and uncover how our Builder communities are utilizing Amazon Bedrock at neighborhood.aws.
— Danilo
