-
Notifications
You must be signed in to change notification settings - Fork 1
Expand file tree
/
Copy pathlmdeploy.sh
More file actions
executable file
·34 lines (30 loc) · 1.29 KB
/
Copy pathlmdeploy.sh
File metadata and controls
executable file
·34 lines (30 loc) · 1.29 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
set -e
mkdir -p logs
export HF_DATASETS_OFFLINE=1
export TRANSFORMERS_OFFLINE=1
LOCAL_MODEL_DIR="your-local-llm-checkpoint-folder"
nohup sh -c "CUDA_VISIBLE_DEVICES=0 python -m lmdeploy serve api_server \
${LOCAL_MODEL_DIR}/Qwen/Qwen2.5-Coder-7B-Instruct \
--model-name Qwen/Qwen2.5-Coder-7B-Instruct \
--server-port 10403 \
--tp 1" > logs/qwen-coder.log 2>&1 &
nohup sh -c "CUDA_VISIBLE_DEVICES=1 python -m lmdeploy serve api_server \
${LOCAL_MODEL_DIR}/google/gemma-2-9b-it \
--model-name google/gemma-2-9b-it \
--server-port 10401 \
--tp 1" > logs/gemma.log 2>&1 &
nohup sh -c "CUDA_VISIBLE_DEVICES=2 python -m lmdeploy serve api_server \
${LOCAL_MODEL_DIR}/mistralai/Ministral-8B-Instruct-2410 \
--model-name mistralai/Ministral-8B-Instruct-2410 \
--server-port 10402 \
--tp 1" > logs/ministral.log 2>&1 &
nohup sh -c "CUDA_VISIBLE_DEVICES=3 python -m lmdeploy serve api_server \
${LOCAL_MODEL_DIR}/Qwen/Qwen2.5-Math-7B-Instruct \
--model-name Qwen/Qwen2.5-Math-7B-Instruct \
--server-port 10404 \
--tp 1" > logs/qwen-math.log 2>&1 &
nohup sh -c "CUDA_VISIBLE_DEVICES=5 python -m lmdeploy serve api_server \
${LOCAL_MODEL_DIR}/ContactDoctor/Bio-Medical-Llama-3-8B \
--model-name ContactDoctor/Bio-Medical-Llama-3-8B \
--server-port 10405 \
--tp 1" > logs/biomedical.log 2>&1 &