Instruction Tuning: GeoSignal

Scientific domain adaptation has two main steps during instruction tuning.

• Instruction tuning with general instruction-tuning data. Here we use Alpaca-GPT4.
• Instruction tuning with restructured domain knowledge, which we call expertise instruction tuning. For K2, we use knowledge-intensive instruction data, GeoSignal.

The following is the illustration of the training domain-specific language model recipe:

• Adapter Model on Huggingface: daven3/k2_it_adapter

For the design of the GeoSignal, we collect knowledge from various data sources, like:

GeoSignal is designed for knowledge-intensive instruction tuning and used for aligning with experts.

The full-version will be upload soon, or email daven for potential research cooperation.