Eval in all-in-one Web UI mmlu

[Katehuuh]

I want to make a eval of benchmarks MMLU, C-Eval and CMMLU as said in Changelog:
[23/09/23] We integrated MMLU, C-Eval and CMMLU benchmarks in this repo. See this example to evaluate your models.

Evaluation

CUDA_VISIBLE_DEVICES=0 python src/evaluate.py \
    --model_name_or_path path_to_llama_model \
    --adapter_name_or_path path_to_checkpoint \
    --template vanilla \
    --finetuning_type lora \
    --task mmlu \
    --split test \
    --lang en \
    --n_shot 5 \
    --batch_size 4

But in Web UI, can we have UI option dropdown of mmlu, ceval and cmmlu?

Not as important given all "official" benchmark but I also suggest add some, start by popular: HumanEval, HellaSwag, ARC, SuperGLUE, GSM8K, TruthfulQA, BIG-Bench, PIQA, Natural Questions, AGIEval, RealToxicity.

[Egber1t]

确实需要一个在web端进行eva加载数据集的下拉栏