updated scripts

llm_toolkit/eval.py

@@ -0,0 +1 @@
+eval_shots.py

scripts/eval-4gpu.sh

@@ -17,8 +17,9 @@ grep MemTotal /proc/meminfo
 
 export BATCH_SIZE=1
 # export START_REPETITION_PENALTY=1.06
+export START_NUM_SHOTS=50
 
-./scripts/eval-model.sh Qwen/Qwen2-72B-Instruct
+#./scripts/eval-model.sh Qwen/Qwen2-72B-Instruct
 
 ./scripts/eval-model.sh shenzhi-wang/Llama3.1-70B-Chinese-Chat
 

scripts/eval-mac.sh

@@ -11,16 +11,16 @@ cat /etc/os-release
 lscpu
 grep MemTotal /proc/meminfo
 
-# pip install torch torchvision torchaudio
-# pip install -r requirements.txt
+pip install torch torchvision torchaudio
+pip install -r requirements.txt
 
-# export START_NUM_SHOTS=50
+export START_NUM_SHOTS=50
 
-./scripts/eval-model.sh internlm/internlm2_5-7b-chat
+#./scripts/eval-model.sh internlm/internlm2_5-7b-chat
 
-./scripts/eval-model.sh Qwen/Qwen2-7B-Instruct
+#./scripts/eval-model.sh Qwen/Qwen2-7B-Instruct
 
-./scripts/eval-model.sh shenzhi-wang/Mistral-7B-v0.3-Chinese-Chat
+#./scripts/eval-model.sh shenzhi-wang/Mistral-7B-v0.3-Chinese-Chat
 
 ./scripts/eval-model.sh shenzhi-wang/Llama3.1-8B-Chinese-Chat
 

scripts/tune-lf-4gpu.sh

@@ -9,8 +9,10 @@ export ORG_NAME=$1
 export MODEL_NAME=$2
 export CHAT_TEMPLATE=$3
 export DATA_PATH=../datasets/mac/mac.tsv
-#export YAML=config/mac_template_4gpu.yaml
-export YAML=config/mac_template_qwen2_72b.yaml
+export YAML=config/mac_template_4gpu.yaml
+#export YAML=config/mac_template_qwen2_72b.yaml
+
+export PYTORCH_CUDA_ALLOC_CONF="expandable_segments:True"
 
 python ../llm_toolkit/setup_lf.py
 llamafactory-cli train config/models/$MODEL_NAME.yaml

scripts/tune-lf.sh

@@ -11,5 +11,7 @@ export CHAT_TEMPLATE=$3
 export DATA_PATH=../datasets/mac/mac.tsv
 export YAML=config/mac_template.yaml
 
+export PYTORCH_CUDA_ALLOC_CONF="expandable_segments:True"
+
 python ../llm_toolkit/setup_lf.py
 llamafactory-cli train config/models/$MODEL_NAME.yaml

scripts/tune-mac-4gpu.sh

@@ -12,8 +12,8 @@ lscpu
 grep MemTotal /proc/meminfo
 
 #pip install -r requirements.txt
-#cd ../LLaMA-Factory && pip install -e .[torch,metrics,vllm] && cd -
+cd ../LLaMA-Factory && pip install -e .[torch,metrics,vllm] && cd -
 
-PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True ./scripts/tune-lf-4gpu.sh Qwen Qwen2-72B-Instruct qwen
+./scripts/tune-lf-4gpu.sh Qwen Qwen2-72B-Instruct qwen
 
 #./scripts/tune-lf-4gpu.sh shenzhi-wang Llama3.1-70B-Chinese-Chat llama3