Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

LLM Inference on LLaMA-2 70B sequence length 2048

384Max Batch Size

CXL-SpecKV + Comp

1.28100.64200299.36Dec 11, 2025
Updated 4d ago

Evaluation Results

MethodLinks
2025.12
384-24
2025.12
192-12
2025.12
128-8
2025.12
48-3
2025.12
16-1