Skip to content

Repository files navigation

CONTOR: Benchmarking Strategies for Completing Ontologies with Plausible Missing Rules

Models uses with QLoRa:

* meta-llama/Llama-2-7b-hf
* meta-llama/Llama-2-7b-chat-hf
* togethercomputer/Llama-2-7B-32K-Instruct
* meta-llama/Llama-2-13b-hf 
* meta-llama/Llama-2-13b-chat-hf 
* mistralai/Mistral-7B-v0.3 
* mistralai/Mistral-7B-Instruct-v0.2 
* lmsys/vicuna-13b-v1.5
* lmsys/vicuna-13b-v1.5-16k
* meta-llama/Meta-Llama-3-8B
* meta-llama/Meta-Llama-3-8B-Instruct

Models uses with unsloth:

* unsloth/mistral-7b-v0.3-bnb-4bit
* unsloth/mistral-7b-instruct-v0.3-bnb-4bit
* unsloth/llama-3-8b-bnb-4bit
* unsloth/llama-3-8b-Instruct-bnb-4bit
* unsloth/Phi-3-medium-4k-instruct-bnb-4bit
* unsloth/gemma-7b-bnb-4bit
* unsloth/gemma-7b-it-bnb-4bit

Examples datasets (untyped/train.json):

{
  "v_sub_concept": "bauxite", 
  "v_super_concept": "mining product", 
  "label": 1, 
  "rule": "body=Geography.Bauxite, head=economy.MiningProduct"
}

Example of a command line:

py main.py --theme all --model_name meta-llama/Llama-2-7b-hf --quantize qlora --prompt 0 --dataset untyped --splited 0 --output ./output

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

1 watching

Forks

Releases

Packages

Contributors

Languages