{"base_model":"Qwen/Qwen3.5-2B","lora_rank":32,"init":"tinker://68826b4d3bd91c3487d3433a3cd5252e/weights/rx_gs1_shared_init","optimizer":{"type":"adamw","learning_rate":4e-05,"beta1":0.9,"beta2":0.95,"eps":1e-08},"loss_fn":"importance_sampling","steps":40,"rollouts_per_step":64,"arms":{"grpo_gs8":{"group_size":8,"groups_per_batch":8},"gs1_kernel":{"group_size":1,"groups_per_batch":64,"window":256,"bandwidth":1.0,"mass_floor":0.001,"phi":"(log rendered prompt length, mean sampler prompt-token logprob), z-scored over window"}},"max_tokens":512,"temperature":1.0,"renderer":"qwen3_5_disable_thinking","data":"openai/gsm8k main train shuffled seed 0 (cookbook Gsm8kDataset), first 2560 prompts for gs1 / 320 for grpo","eval":"greedy, max_tokens 512, 1319 test items"}