Commit ac238727 authored by Lianmin Zheng's avatar Lianmin Zheng
Browse files

Support penalty in overlap mode; return logprob with chunked prefill; improve...


Support penalty in overlap mode; return logprob with chunked prefill; improve benchmark scripts (#3988)
Co-authored-by: default avatarSangBin Cho <rkooo567@gmail.com>
Co-authored-by: default avatardhou-xai <dhou@x.ai>
Co-authored-by: default avatarHanming Lu <hanming_lu@berkeley.edu>
parent 0194948f
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
#!/bin/bash #!/bin/bash
# Check if sudo is available
if command -v sudo >/dev/null 2>&1; then
sudo apt-get update
sudo apt-get install -y lsof
else
apt-get update
apt-get install -y lsof
fi
# Show current GPU status # Show current GPU status
nvidia-smi nvidia-smi
...@@ -20,6 +11,14 @@ kill -9 $(ps aux | grep 'sglang.data_parallel' | grep -v 'grep' | awk '{print $2 ...@@ -20,6 +11,14 @@ kill -9 $(ps aux | grep 'sglang.data_parallel' | grep -v 'grep' | awk '{print $2
# Clean all GPU processes if any argument is provided # Clean all GPU processes if any argument is provided
if [ $# -gt 0 ]; then if [ $# -gt 0 ]; then
# Check if sudo is available
if command -v sudo >/dev/null 2>&1; then
sudo apt-get update
sudo apt-get install -y lsof
else
apt-get update
apt-get install -y lsof
fi
kill -9 $(nvidia-smi | sed -n '/Processes:/,$p' | grep " [0-9]" | awk '{print $5}') 2>/dev/null kill -9 $(nvidia-smi | sed -n '/Processes:/,$p' | grep " [0-9]" | awk '{print $5}') 2>/dev/null
lsof /dev/nvidia* | awk '{print $2}' | xargs kill -9 2>/dev/null lsof /dev/nvidia* | awk '{print $2}' | xargs kill -9 2>/dev/null
fi fi
......
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment