AI · អត្ថបទ
AWS Adds Full-Weight Training to Serverless SageMaker Model Customization
Unite.AI ·

មើលត្រួសៗ
Amazon Web Services announced full fine-tuning support for its serverless SageMaker AI model customization service. In a coding test, the method reached 90.0% semantic accuracy, but AWS warned that overly aggressive settings can erase general knowledge.
ចំណុចសំខាន់ៗ
- Full fine-tuning updates every weight in the model, while LoRA only trains small added adapter layers.
- In AWS's test, full fine-tuning beat LoRA and the base model on every quality measure shown, with the biggest gaps on BLEU and ROUGE-2.
- Full fine-tuning did not win everywhere: on MATH exact match, both methods scored below the base model.
- A learning rate of 5e-5 or higher caused sharp drops, so AWS suggested 2e-5 for large datasets.
ហេតុអ្វីវាសំខាន់
Deeper customization can improve results on specific tasks, but it costs more compute and can make a model forget general knowledge.
ប្រភពដើម
AWS, 서버리스 SageMaker 맞춤화에 전체 가중치 학습 도입
Unite.AI · Nov 19, 09:39 KST
Read on Unite.AI ↗AI summarized the original article. Check the original for accuracy. Copyright belongs to Unite.AI.
ការផ្សាយផ្សេងទៀត
Amazon Web Services · ចំណាត់ថ្នាក់ និងព័ត៌មានបន្ថែម
បន្ថែមពីវិស័យនេះ
- Designers Say Ssense's Website Shows Their Clothes Inaccurately
Designer Reese Cooper noticed his RCOS puffer coat looked distorted on the luxury retailer Ssense's website, with a chest hook drawn far too large. Ssense said some site images came from a "virtual try-on" program, though it would not say whether that program uses AI.
The New York Times · Nov 21, 06:16 KST - Korean Firms Build Three-Way AI Push: Anthropic Stake, HBM Chips, Claude Rollout
Korean companies are expanding their roles in the global AI market around Anthropic's IPO. SK Telecom earned large paper gains from an early investment, while Samsung Electronics and SK hynix are strengthening chip supply ties. Samsung SDS and LG CNS are bringing Anthropic's Claude model into the business AI market.
뉴스웨이 · Nov 21, 06:10 KST - MIT and IBM Team Probes Gaps in Multimodal AI Understanding
MIT and IBM researchers are studying where today's multimodal AI models fall short, even when they handle text, images, and sound fluently. Their team, co-led by MIT's Jim Glass, builds new benchmarks and training methods that need less labeled data. This Q&A walks through how the project started and how it has evolved.
MIT News · Nov 21, 06:00 KST - Leaky AI Supply Chain Exposes Secret User Data at Alarming Rate
A research test sent nearly two million prompts with fake but working secrets through a network of AI services to see where they ended up. Many of the secrets were stolen, and attackers used them to reach cloud accounts. The results point to weak protection of customer data along the chain.
Ars Technica · Nov 21, 05:59 KST