openvino
[NPUW] Share kvcache between prefill and generate when chunking is enabled
#32642
Merged

Commits
  • Introduce lazy memory allocation for ireq's I/O
    smirnov-alexey committed 314 days ago
  • Fix no tensor being present in the storage
    smirnov-alexey committed 314 days ago
  • Merge branch 'master' of https://github.com/openvinotoolkit/openvino into as/npuw_lazy_io_alloc
    smirnov-alexey committed 307 days ago
  • Address review comments
    smirnov-alexey committed 306 days ago
  • Merge branch 'master' of https://github.com/openvinotoolkit/openvino into as/npuw_lazy_io_alloc
    smirnov-alexey committed 303 days ago
  • Merge branch 'master' of https://github.com/openvinotoolkit/openvino into as/npuw_share_kvcache
    smirnov-alexey committed 285 days ago
  • Copy Xiong's changes
    smirnov-alexey committed 285 days ago
  • Remove copy
    smirnov-alexey committed 282 days ago
  • WIP
    smirnov-alexey committed 281 days ago
  • Fix concurrency issue with iterator invalidation
    smirnov-alexey committed 280 days ago
  • Merge branch 'master' of https://github.com/openvinotoolkit/openvino into as/npuw_lazy_io_alloc
    smirnov-alexey committed 280 days ago
  • Refactoring
    smirnov-alexey committed 280 days ago
  • Fix merge
    smirnov-alexey committed 280 days ago
  • Protect get_tensor by mutex
    smirnov-alexey committed 280 days ago
  • Merge branch 'as/npuw_lazy_io_alloc' of https://github.com/smirnov-alexey/openvino into as/npuw_share_kvcache
    smirnov-alexey committed 280 days ago
  • Disable kv cache sharing when one of the models is transposed
    smirnov-alexey committed 280 days ago
  • Fix strides
    smirnov-alexey committed 278 days ago
  • Handle strided tensors - copy on host
    smirnov-alexey committed 275 days ago
  • Merge branch 'master' of https://github.com/openvinotoolkit/openvino into as/npuw_share_kvcache
    smirnov-alexey committed 275 days ago
  • Handle strided tensors in pyramid attention
    smirnov-alexey committed 275 days ago
  • Address review comments
    smirnov-alexey committed 274 days ago
  • Merge branch 'as/npuw_lazy_io_alloc' of https://github.com/smirnov-alexey/openvino into as/npuw_share_kvcache
    smirnov-alexey committed 274 days ago
  • Address review comments
    smirnov-alexey committed 274 days ago
  • Merge branch 'master' into as/npuw_lazy_io_alloc
    smirnov-alexey committed 274 days ago
  • Merge branch 'master' into as/npuw_share_kvcache
    smirnov-alexey committed 274 days ago
  • Fix shape
    smirnov-alexey committed 274 days ago
  • Merge branch 'master' of https://github.com/openvinotoolkit/openvino into as/npuw_lazy_io_alloc
    smirnov-alexey committed 261 days ago
  • Remove is_io()
    smirnov-alexey committed 261 days ago
  • Merge branch 'as/npuw_lazy_io_alloc' of https://github.com/smirnov-alexey/openvino into as/npuw_share_kvcache
    smirnov-alexey committed 261 days ago
  • Move kv cache copy to second token time
    smirnov-alexey committed 261 days ago
  • Fix empty tensors for pyramid attention
    smirnov-alexey committed 261 days ago
  • Merge branch 'as/npuw_share_kvcache' of https://github.com/smirnov-alexey/openvino into as/npuw_share_kvcache
    smirnov-alexey committed 261 days ago
  • Merge branch 'master' into as/npuw_share_kvcache
    smirnov-alexey committed 258 days ago
  • Align code with latest pyramid changes merged
    smirnov-alexey committed 258 days ago
Loading