kv-cells : track min/max used cells and per-sequence positions #13808
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
ggerganov
force-pushed
the
gg/kv-cells-track-min-max
branch
from
dc6b26d to
9bf13fd
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
cont #13706
The
llama_kv_cells_unifiednow tracks 2 new min/max quantities:The min/max indices of used cells
Mainly needed for determining the range of KV cells
[0, n_kv)which are considered during the attention computation with unified KV cache (a.k.a. the oldcell_max()).The min/max positions for each sequence currently present in the KV cache
Will be used in kv-cache : refactor + add llama_memory_state_i #13746 to improve the
find_slot()logic in SWA cases.To amortize the cost, we utilize
std::set.