devflow / analysis_outputs /T64 /task1_kv_cache.txt
bhsinghgrid's picture
Update app/inference + ablation task outputs
27f26fd verified
TASK 1 — KV CACHE BENCHMARK
========================================
has_generate_cached=True
memory_profile=N/A (CPU/MPS)
src_len standard(s) cached(s) speedup encoder%
16 7.223 4.472 1.62x 41.7%
32 6.365 4.063 1.57x 41.3%
64 10.223 6.790 1.51x 41.9%
Saved graphs:
- task1_time_comparison.png
- task1_speedup.png
- task1_encoder_cost.png