test gemma2

scripts/test-mgtv.sh

@@ -13,23 +13,28 @@ cat /etc/os-release
 lscpu
 grep MemTotal /proc/meminfo
 
-pip install transformers==4.41.2
+#pip install transformers==4.41.2
 
 export LOAD_IN_4BIT=false
-export USING_LLAMA_FACTORY=true
+#export USING_LLAMA_FACTORY=true
+export USING_LLAMA_FACTORY=false
 
 #export MODEL_NAME=internlm/internlm2_5-7b-chat-1m
 # export ADAPTER_NAME_OR_PATH=inflaton-ai/InternLM_2_5-7b_LoRA-Adapter
 #export ADAPTER_NAME_OR_PATH=llama-factory/saves/internlm2_5_7b/lora/sft_bf16_p2_full_r3/checkpoint-140
 
-export MODEL_NAME=shenzhi-wang/Llama3-8B-Chinese-Chat
-export ADAPTER_NAME_OR_PATH=llama-factory/saves/llama3-8b/lora/sft_bf16_p1_full_r4/checkpoint-140
-export USING_P1_PROMPT_TEMPLATE=true
+export MODEL_NAME=google/gemma-2-9b-it
+export ADAPTER_NAME_OR_PATH=outputs/checkpoint-6000
+export USING_P1_PROMPT_TEMPLATE=false
+
+#export MODEL_NAME=shenzhi-wang/Llama3-8B-Chinese-Chat
+#export ADAPTER_NAME_OR_PATH=llama-factory/saves/llama3-8b/lora/sft_bf16_p1_full_r4/checkpoint-140
+#export USING_P1_PROMPT_TEMPLATE=true
 
 export LOGICAL_REASONING_DATA_PATH=datasets/mgtv
 
 export TEST_DATA=test_b
-export LOGICAL_REASONING_RESULTS_PATH=results/$TEST_DATA-results_r4.csv
+export LOGICAL_REASONING_RESULTS_PATH=results/$TEST_DATA-results_r5.csv
 
 echo "Eval $MODEL_NAME with $ADAPTER_NAME_OR_PATH"
 python llm_toolkit/eval_logical_reasoning.py