TensorRT-LLMs/tests/integration/defs/accuracy/references/longbench_v2.yaml
Fanrong Li 2f526583fb
[None][chore] Move the rocketkv e2e test to post-merge (#9768)
Signed-off-by: Fanrong Li <23290157+lfr-0531@users.noreply.github.com>
2025-12-08 13:22:16 +08:00

13 lines
289 B
YAML

DeepSeek-R1-0528:
- quant_algo: FP8_BLOCK_SCALES
kv_cache_quant_algo: FP8
spec_dec_algo: MTP
accuracy: 52.093
- quant_algo: NVFP4
kv_cache_quant_algo: FP8
spec_dec_algo: MTP
accuracy: 52.093
meta-llama/Llama-3.1-8B-Instruct:
- accuracy: 26.00
sigma: 25.8