{"base_model":"Qwen/Qwen3.5-2B","renderer":"qwen3_5","lora_rank":32,"init":"tinker://09950f53d8f9d5a7e24607f09804e8c3/weights/rx-shared-init","loss":"importance_sampling","optimizer":"AdamParams(lr=4e-5,beta1=0.9,beta2=0.95,eps=1e-8)","max_tokens":512,"temperature":1.0,"rollouts_per_step":64,"steps_planned":40,"grpo":{"group_size":8,"groups":8},"gs1":{"group_size":1,"groups":64,"window":256,"bandwidth":0.5,"phi":"(log prompt tokens, sampler mean prompt-token logprob), z-scored over window","fallback_mass":0.001},"train_order":"openai/gsm8k main train shuffle(seed=0) first 2560"}