Skip to content

INT8 KV Experiments #3

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 5 commits into
base: main
Choose a base branch
from
Open

INT8 KV Experiments #3

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
c6accde
Update phi3 example: increase max_length to 6K tokens and fix API com…
anujj Jun 9, 2025
f663a5b
application changes
anujj Jun 9, 2025
f73deba
Add int8 KV cache quantization support
anujj Jun 16, 2025
7c875cc
revert exp changes
anujj Jun 23, 2025
b7d8700
exp changes
anujj Jun 23, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 10 additions & 3 deletions src/models/kv_cache.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -175,8 +175,14 @@ DefaultKeyValueCache::DefaultKeyValueCache(State& state)
}

// Set the size after empty_past_ has been created with 0 for this field
if (past_present_share_buffer_)
if (state.model_.p_device_->GetType() == DeviceType::NvTensorRtRtx &&
model_.config_->model.decoder.sliding_window.has_value() &&
model_.config_->model.decoder.sliding_window->window_size > 0) {
shape_[2] = std::min(state_.params_->search.max_length,
model_.config_->model.decoder.sliding_window->window_size);
} else if (past_present_share_buffer_) {
shape_[2] = state_.params_->search.max_length;
}

try {
for (int i = 0; i < layer_count_ * 2; ++i) {
Expand Down Expand Up @@ -422,12 +428,13 @@ std::unique_ptr<KeyValueCache> CreateKeyValueCache(State& state) {
return nullptr;
}

if (state.model_.config_->model.decoder.sliding_window &&
if (state.model_.p_device_->GetType() != DeviceType::NvTensorRtRtx &&
state.model_.config_->model.decoder.sliding_window &&
state.model_.config_->model.decoder.sliding_window->slide_key_value_cache) {
return std::make_unique<WindowedKeyValueCache>(state);
}

return std::make_unique<DefaultKeyValueCache>(state);
}

} // namespace Generators
} // namespace Generators
Loading