A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 9, 20:45 UTC
[ROCm][CI] Lower Acceptance Len Threshold For test_draft_model_quantization (#32731)
Signed-off-by: Micah Williamson <micah.williamson@amd.com>
Micah Williamson committed
Jan 21, 2026 at 23:47 UTC
019e2c3b7ca799e4497a67eb7e49edea361cb90e
1 file changed
+1
-1
tests/v1/e2e/test_spec_decode.py
+1
-1
@@ -808,7 +808,7 @@ def some_high_acceptance_metrics() -> dict:
808
return {
809
"sampling_config": greedy_sampling(),
810
"num_speculative_tokens": 3,
811
- "expected_acceptance_len": 2.90 + 1,
811
+ "expected_acceptance_len": 2.8 + 1,
812
"expected_acceptance_rate": 0.90,
813
}
814