Skip to content
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

[Bugfix] Fix value unpack error of simple connector for KVCache transfer. #11058

Merged
merged 3 commits into from
Dec 12, 2024
Merged

[Bugfix] Fix value unpack error of simple connector for KVCache transfer. #11058

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
f0b13d4
Fix value unpack error of simple connector.
ShangmingCai Dec 10, 2024
ecfb7f0
retrigger ci
ShangmingCai Dec 10, 2024
b8fc9aa
Fix typo.
ShangmingCai Dec 11, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 6 additions & 2 deletions vllm/distributed/kv_transfer/kv_connector/simple_connector.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -118,6 +118,12 @@ def send_kv_caches_and_hidden_states(
start_layer = model_executable.model.start_layer
end_layer = model_executable.model.end_layer

model_config = model_executable.model.config
num_heads = model_config.num_key_value_heads
hidden_size = model_config.hidden_size
num_attention_heads = model_config.num_attention_heads
head_size = int(hidden_size / num_attention_heads)

# query_lens contains new KV caches that are added to vLLM.
# so we will send them to decode instance
# FIXME(Kuntai): This assume that all requests are prefill.
Expand All @@ -131,8 +137,6 @@ def send_kv_caches_and_hidden_states(
for layer_id in range(start_layer, end_layer):
kv_cache = kv_caches[layer_id - start_layer]

_, _, num_heads, head_size = kv_cache[0].shape

key_cache = kv_cache[0].reshape(-1, num_heads, head_size)
value_cache = kv_cache[1].reshape(-1, num_heads, head_size)

Expand Down
Loading