Finetuning-LLMs

Fine tuning different Large Language Models (LLMs) using variety of techniques available in OSS world

FT Mistral 7B on Medicine data: MedMCQA dataset(QA pairs) to achieve ~45% accuracy in answering question
FT Openllama 3b v2 on 'Red dot product description' to generate product description for 'suggested' products.
FT Mistral 7B on Instruction dataset : MosaicML to generate possible question for a given description text
FT Unsloth Mistral 7B on Alpaca dataset. Unsloth provides quantized models which load faster and consume less memory during training and inference.

Name		Name	Last commit message	Last commit date
AIMKS_GC_done_of_Finetuning_Mistral_7B_mosaicml_15124.ipynb		AIMKS_GC_done_of_Finetuning_Mistral_7B_mosaicml_15124.ipynb
LICENSE		LICENSE
README.md		README.md
done_GC_Unsloth_Alpaca_+_Mistral_7b_full_example_15_1_24.ipynb		done_GC_Unsloth_Alpaca_+_Mistral_7b_full_example_15_1_24.ipynb
gc-kag-gld-mistral-7b-ft-medmcqa-251223-ipynb-v5.ipynb		gc-kag-gld-mistral-7b-ft-medmcqa-251223-ipynb-v5.ipynb
kag-gld-Falcon7B-mental-health-heliosbrah-25124.ipynb		kag-gld-Falcon7B-mental-health-heliosbrah-25124.ipynb
ok-kag-ft-openllama3b-v2-product-desc-11124-v4.ipynb		ok-kag-ft-openllama3b-v2-product-desc-11124-v4.ipynb

Navigation Menu