Skip to content

[None][chore] Remove unused token_range and is_last_slice parameters from _create_kv_slice #16083

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

Merged
Tabrizian merged 2 commits into from
Jul 13, 2026
Merged

[None][chore] Remove unused token_range and is_last_slice parameters from _create_kv_slice #16083

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
5ff84e8
[None][chore] Remove unused token_range parameter from _create_kv_slice
Tabrizian Jul 7, 2026
ce9b56b
[None][chore] Remove unused is_last_slice parameter from _create_kv_s…
Tabrizian Jul 7, 2026
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
12 changes: 4 additions & 8 deletions tensorrt_llm/_torch/disaggregation/transceiver.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -173,12 +173,7 @@ def __enter__(self):
def __exit__(self, _exc_type, _exc_val, _exc_tb):
self.shutdown()

def _create_kv_slice(
self,
req: LlmRequest,
token_range: Optional[TokenRange] = None,
is_last_slice: bool = True,
) -> KVSlice:
def _create_kv_slice(self, req: LlmRequest) -> KVSlice:
adapter = self._reuse_adapter
tpb = adapter.tokens_per_block
assert self._page_table is not None
Expand All @@ -191,7 +186,8 @@ def _create_kv_slice(
else [0] * len(layer_groups)
)

if token_range is None and req.prompt_len > 0:
token_range = None
if req.prompt_len > 0:
# Align with KV cache allocation (prepare_disagg_gen_init /
# _get_context_bytes), which reserves prompt_len +
# num_extra_kv_tokens slots for speculative decoding methods
Expand Down Expand Up @@ -242,7 +238,7 @@ def _create_kv_slice(
mamba_state_index = self._kv_cache_manager.mamba_cache_index[req.py_request_id]

return KVSlice(
is_last_slice=is_last_slice,
is_last_slice=True,
block_ids_per_layer_groups=groups,
mamba_state_index=mamba_state_index,
token_range=token_range,
Expand Down
11 changes: 0 additions & 11 deletions tests/unittest/disaggregated/test_cache_reuse_adapter.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -342,17 +342,6 @@ def test_defaults_to_prompt_len_when_no_extra(self):
assert kv_slice.token_range is not None
assert (kv_slice.token_range.start, kv_slice.token_range.end) == (0, prompt_len)

def test_respects_explicit_token_range(self):
prompt_len = 17
transceiver, req = _build_transceiver_for_kv_slice(
num_extra_kv_tokens=7, prompt_len=prompt_len
)
explicit = TokenRange(start=0, end=8)

kv_slice = transceiver._create_kv_slice(req, token_range=explicit)

assert kv_slice.token_range is explicit


# ---------------------------------------------------------------------------
# CacheReuseAdapter.get_cached_token_count_per_layer_group: SWA clamp.
Expand Down
Loading