[BugFix][Spec Decode] Remove obsolete sequence length update (#15105)
### What this PR does / why we need it? This PR removes the unused `_ascend_update_seq_lens` method and its invocation in `_run_model` within `AscendAutoRegressiveSpeculator`. The method runs after the draft model execution has completed, and the generated `seq_len_list` is not consumed by the subsequent speculative decoding flow. Sequence lengths needed by draft attention are already prepared through the existing attention metadata update path, so this additional update is unnecessary. ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? - `python -m py_compile vllm_ascend/worker/v2/spec_decode/autoregressive/speculator.py` - `git diff --check` Signed-off-by: jiajinzhu2 <jiajinzhu@huawei.com>
J
jiajinzhu2 committed
23c8bf712c5c2963e12c8488551bcc7837f7fca7
Parent: 8091873
Committed by GitHub <noreply@github.com>
on 8/27/2026, 10:24:50 AM