Correct alignment in the seq_len diagram. (#5592)

Co-authored-by: Liqian Chen <liqian.chen@deeplang.ai>

Correct alignment in the seq_len diagram. (#5592)
Co-authored-by: Liqian Chen <liqian.chen@deeplang.ai>
9e74d9d0 · Charles Riggins · GitHub · 9333fb8e · 9e74d9d0
Unverified Commit 9e74d9d0 authored Jun 18, 2024 by Charles Riggins Committed by GitHub Jun 17, 2024
Show whitespace changes
Inline Side-by-side

Showing with 1 addition and 1 deletion

vllm/attention/backends/flash_attn.py vllm/attention/backends/flash_attn.py +1 -1

No files found.
--- a/vllm/attention/backends/flash_attn.py
+++ b/vllm/attention/backends/flash_attn.py
@@ -83,7 +83,7 @@ class FlashAttentionMetadata(AttentionMetadata):
    # |---------------- N iteration ---------------------|
    # |- tokenA -|......................|-- newTokens ---|
    # |---------- context_len ----------|
-    # |-------------------- seq_len ----------------------|
+    # |-------------------- seq_len ---------------------|
    #                                   |-- query_len ---|

    # Maximum query length in the batch. None for decoding.