README.md

metadata

license: apache-2.0
datasets:
  - spider
metrics:
  - accuracy
model-index:
  - name: lever-spider-codex
    results:
      - task:
          type: code generation
        dataset:
          type: spider
          name: spider (text-to-sql)
        metrics:
          - type: accuracy
            value: 81.9
            verified: false

LEVER for Codex on Spider

Basic usage: Load the model and tokenizer as follows:

tokenizer = T5Tokenizer.from_pretrained("Yale-LILY/lever-spider-codex")
model = T5ForConditionalGeneration.from_pretrained("Yale-LILY/lever-spider-codex")