v0.9.0

What's Changed

• Introduce category_key for flexeval_reward by @ryokan0123 in #107
• Move vllm import inside function to avoid ImportError by @ryokan0123 in #106
• Set self.model.config.pad_token_id from tokenizer if not set in SequenceClassificationRewardModel by @ryokan0123 in #108
• Add new StringProcessors by @teruaki-o in #110
• Replace Deprecated parameter max_tokens with max_completion_tokens for OpenAI API by @junya-takayama in #112
• Cache jinja2 template by @butsugiri in #113
• Read prefix in perplexity evaluation by @ryokan0123 in #114
• Quit sharing StringProcessor between lm_output and reference by @ryokan0123 in #115
• Add Correlation metric class by @ryokan0123 in #116
• Remove NoopNormalizer by @ryokan0123 in #117

Full Changelog: v0.8.1...v0.9.0