BeastxD
/

text2cypher_lora_v6_shaped

Reinforcement Learning

Model card Files Files and versions

text2cypher_lora_v6_shaped / tokenizer.json

Commit History

Aug 26

04:18
v6 shaped-reward GRPO (782 steps, ckpt-600) -- 56.81% on 2024v1, +1.18pp vs v7, p=0.089 (not significant)

BeastxD

Verified signature
72b6c7c