Qwen2.5-Gutenberg-Doppel-32B
Qwen/Qwen2.5-32B-Instruct finetuned on jondurbin/gutenberg-dpo-v0.1 and nbeerbower/gutenberg2-dpo.
Method
ORPO tuned with 2x A100 for 1.25 epochs.
Qwen/Qwen2.5-32B-Instruct finetuned on jondurbin/gutenberg-dpo-v0.1 and nbeerbower/gutenberg2-dpo.
ORPO tuned with 2x A100 for 1.25 epochs.