@darryl-ruggles.cloud#StrandsAgents #BidiAgents #AmazonNova #OpenTelemetry #Agents
Voice agents get less attention than most areas of agent development. I have built a few apps with them, and talking to a speech-to-speech agent is a different experience. Strands just made Bidi Agents GA, with
@promptfoundry.bsky.socialA Taiwan retail platform customized Amazon Nova Lite through fine-tuning and prompt optimization…
#GenerativeAI #PromptEngineering #AmazonNova #RetailTech
https://aws.amazon.com/blogs/machine-learning/how-uniopen-customized-amazon-nova-to-their-retail-moderation-policies-for-production-deployment/
@sergiocuellar.bsky.socialAmazon Nova's second post on data preparation for supervised fine-tuning (SFT) dives into advanced strategies like data subset selection, augmentation, and mixing for optimal model training. Measure first before adding more data! #AmazonNova #SFT #DataPreparation
@globalskiatlas.bsky.socialAsked the skiing AI which Fast and Furious movie is best. Nova Pro swerved straight back to the mountains with a list of ski resorts, and Val Thorens got linked to its atlas page. #AmazonNova #buildinpublic
@globalskiatlas.bsky.socialGlobal Ski Atlas has a skiing AI now. Ask for a resort or the biggest in a state, and it answers from the atlas index with wiki and map links. Anything else goes to Nova Pro on AWS Bedrock. globalskiatlas.com/skiing-ai.html #AmazonNova #buildinpublic #skiing
@sergiocuellar.bsky.social"Designing custom rewards for Amazon Nova models in multi-turn RL tasks is crucial for optimal performance. Learn how to avoid pitfalls and optimize rewards for effective training. 🤖🎓 #AmazonNova #RL #CustomRewards"