New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

KV cache size calculation is inaccurate. #11

Open

future-xy opened this issue May 28, 2024 · 1 comment

Assignees

future-xy commented May 28, 2024

open-moe-llm-leaderboard/src/backend/hflm_with_measurement.py

Line 436 in 54dbe86

per_token_kv_size = 2 * n_layers * d_model * precision_bytes

This code doesn't consider "GQA" and "MQA".

future-xy assigned future-xy and SecretSettler and unassigned future-xy

Author

future-xy commented May 28, 2024

The correct way to calculate is:

d_head = d_model / num_attention_heads
per_token_kv_size = 2 * n_layers * d_head * num_key_value_heads * precision_bytes

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment