Skip to content

Instantly share code, notes, and snippets.

View binzhango's full-sized avatar
💭
I may be slow to respond.

Bin Zhang binzhango

💭
I may be slow to respond.
View GitHub Profile
@binzhango
binzhango / vllm.sh
Last active December 12, 2025 13:27
export VLLM_USE_FLASH_ATTENTION=0
export VLLM_FLASH_ATTN_VERSION=0
export VLLM_ATTENTION_BACKEND=TORCH_SDPA
export VLLM_DISABLE_FLASHINFER_PREFILL=1
vllm serve "Qwen/Qwen3-1.7B-Base" --gpu-memory-utilization 0.80 --max-num-batched-tokens 2048 --max-num-seqs 1
# keytool -exportcert -rfc -alias <truststore-alias> -keystore truststore.jks -file truststore.pem
# keytool -importkeystore -srckeystore keystore.jks -destkeystore keystore.p12 -srcstoretype JKS -deststoretype PKCS12 -srcalias <keystore-alias> -deststorepass <password> -srcstorepass <password>
# openssl pkcs12 -in keystore.p12 -nodes -nocerts -out key.pem -passin pass:<password>
# openssl pkcs12 -in keystore.p12 -nokeys -out cert.pem -passin pass:<password>
import jks
# Path to your JKS file and the keystore password
keystore_path = 'path/to/your/keystore.jks'