SciGLM: Training Scientific Language Models with Self-Reflective Instruction Annotation and Tuning

📃 [SciGLM] [GitHub]

SciGLM is a suite of scientific language models able to conduct college-level scientific reasoning. Central to our approach is a novel self-reflective instruction annotation framework to address the data scarcity challenge in the science domain. This framework leverages existing LLMs to generate step-by-step reasoning for unlabelled scientific questions, followed by a process of self-reflective critic-and-revise. Applying this framework, we curated SciInstruct, a diverse and high-quality dataset encompassing mathematics, physics, chemistry, and formal proofs.

SciInstruct
Training
Leaderboard
Citation

SciInstruct

We construct the SciInstruct as follows:

Subject	Math	Physics&Chemistry	Formal Proofs (Lean)	Total
# Number	89,934	42,034	40,248	172,216

We will release our data and model for public use.

Training

Fine-tuning

To train the 6B model, run:

LR=3e-6
WR=0.0
LST=linear
epoch=2
MASTER_PORT=$(shuf -n 1 -i 10000-65535)

deepspeed --include="localhost:0,1,2,3" --master_port $MASTER_PORT main.py \
    --deepspeed /ChatGLM2-6B-main/ptuning/deepspeed.json \
    --do_train \
    --train_file SciInstruct.json \
    --eval_steps 200 \
    --prompt_column content \
    --response_column summary \
    --overwrite_cache \
    --model_name_or_path ChatGLM3-6b-Base \
    --output_dir ./output/SciGLM-LR$LR-WR$WR-$LST-epoch$epoch \
    --overwrite_output_dir \
    --max_source_length 128 \
    --max_target_length 512 \
    --per_device_train_batch_size 4 \
    --per_device_eval_batch_size 4 \
    --gradient_accumulation_steps 2 \
    --predict_with_generate \
    --num_train_epochs $epoch \
    --logging_steps 20 \
    --save_strategy epoch \
    --learning_rate $LR \
    --lr_scheduler_type $LST \
    --warmup_ratio $WR \
    --fp16

Leaderboard

Results on Scientific Reasonings tasks

Results on Mathematical Reasoning

Citation

If you find our work helpful, please kindly cite our paper:

@misc{zhang2024sciglm,
      title={SciGLM: Training Scientific Language Models with Self-Reflective Instruction Annotation and Tuning}, 
      author={Dan Zhang and Ziniu Hu and Sining Zhoubian and Zhengxiao Du and Kaiyu Yang and Zihan Wang and Yisong Yue and Yuxiao Dong and Jie Tang},
      year={2024},
      eprint={2401.07950},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}