BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Sabre//Sabre VObject 4.5.8//EN
CALSCALE:GREGORIAN
LAST-MODIFIED:20241016T223651Z
BEGIN:VTIMEZONE
TZID:America/Los_Angeles
X-MICROSOFT-CDO-TZID:13
BEGIN:STANDARD
DTSTART:20231105T090000
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
END:STANDARD
BEGIN:STANDARD
DTSTART:20241103T090000
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20240310T100000
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:90030c25-9d89-4e3c-82f8-ef05200d73db
DTSTAMP:20241016T223651Z
SUMMARY:AI Model Management
DTSTART;TZID=America/Los_Angeles:20240925T131500
DTEND;TZID=America/Los_Angeles:20240925T141500
DESCRIPTION:https://www.w3.org/events/meetings/90030c25-9d89-4e3c-82f8-ef05
 200d73db/\n\nAI models can be executed on the client web platform and can 
 add significant functionality to web applications.   However\, they can al
 so be quite large\, requiring significant resources to download and store.
   Download and compilation latencies can potentially impact the user exper
 ience.\n\nThis breakout will discuss ways in which these issues can be mit
 igated.   Possible topics include the following.   \n\n- Background model 
 download and compilation.\n- Caching strategies\, including potential cros
 s-site caching mechanisms with privacy-preserving mitigations\n- Model nam
 ing and versioning\, allowing for model substitution when useful\n- Access
  to both downloadable and pre-installed models with a common interface\n- 
 Storage deduplication\n- Model representation independence\n- API independ
 ence (e.g. sharing models between WebNN and WebGPU implementations)\n- Off
 line usage\, including interaction with PWAs.\n- Common models are lower p
 rivacy risks \n\nNote: this is both an AI topic and a Storage topic.  Inpu
 t from both communities would be useful and is encouraged!\n\nThere were s
 ome [related presentations](https://github.com/webmachinelearning/hybrid-a
 i/tree/main/presentations) on this topic in the WebML IG.\n\nSee:\n\n- [Re
 po](https://github.com/webmachinelearning/hybrid-ai/) - Please direct foll
 owup there\, and to the WebML WG\n\nAgenda\n\n**Chairs:**\nMichael McCool\
 n\n**Description:**\nAI models can be executed on the client web platform 
 and can add significant functionality to web applications.   However\, the
 y can also be quite large\, requiring significant resources to download an
 d store.  Download and compilation latencies can potentially impact the us
 er experience.\n\nThis breakout will discuss ways in which these issues ca
 n be mitigated.   Possible topics include the following.   \n\n- Backgroun
 d model download and compilation.\n- Caching strategies\, including potent
 ial cross-site caching mechanisms with privacy-preserving mitigations\n- M
 odel naming and versioning\, allowing for model substitution when useful\n
 - Access to both downloadable and pre-installed models with a common inter
 face\n- Storage deduplication\n- Model representation independence\n- API 
 independence (e.g. sharing models between WebNN and WebGPU implementations
 )\n- Offline usage\, including interaction with PWAs.\n- Common models are
  lower privacy risks \n\nNote: this is both an AI topic and a Storage topi
 c.  Input from both communities would be useful and is encouraged!\n\nTher
 e were some [related presentations](https://github.com/webmachinelearning/
 hybrid-ai/tree/main/presentations) on this topic in the WebML IG.\n\nSee:\
 n\n- [Repo](https://github.com/webmachinelearning/hybrid-ai/) - Please dir
 ect followup there\, and to the WebML WG\n\n**Goal(s):**\nPrioritize issue
 s\, discuss highest priority issues\, define follow-up actions if possible
 .\n\n\n**Agenda:**\n1. Review list of issues and add or refine any if nece
 ssary (5m)\n2. Prioritize issues\, identify shortlist for discussion (10m)
 \n3. Discuss potential solutions to high-priority issues (approx 15m each)
 \n      - Expand explanation of each issue\, identify stakeholders\n      
 - Discuss possible resolutions \n      - Define followup actions and colla
 borations\n\n**Materials:**\n- [slides](https://www.w3.org/2024/Talks/TPAC
 /breakouts/ai-model-management.pdf)\n- [minutes](https://www.w3.org/2024/0
 9/25-ai-model-minutes.html)\n- [Session proposal on GitHub](https://github
 .com/w3c/tpac2024-breakouts/issues/15)\n\n**Track(s):**\n- AI
STATUS:CONFIRMED
CREATED:20240916T214850Z
LAST-MODIFIED:20241016T223651Z
SEQUENCE:1
ORGANIZER;CN=W3C Calendar;PARTSTAT=ACCEPTED;ROLE=NON-PARTICIPANT:mailto:nor
 eply@w3.org
LOCATION:2 Ballroom Level - California A
CATEGORIES:TPAC 2024,Breakout Sessions
END:VEVENT
END:VCALENDAR