{ "seed": 297, "device": "cuda", "math_check": { "unsafe_growth_slope": -1.9995359534972887, "safe_variation_slope": 0.00046404650271114774, "safe_limit_error_at_1e-5": 1.2237717207067422e-05, "unsafe_norm_ratio": 99331043.78967242, "safe_norm_ratio": 0.9933104378967241 }, "experiment": { "unsafe": { "val_mse": 799632392192.0, "max_output_norm": 18367922.0, "near_pole_output_norm": 53044752.0, "max_grad_norm": 0.0, "nan_steps": 0, "parameters": 1344 }, "clip": { "val_mse": 133724012544.0, "max_output_norm": 20492250.0, "near_pole_output_norm": 22388642.0, "max_grad_norm": 255317835776.0, "nan_steps": 0, "parameters": 1344 }, "safe": { "val_mse": 0.022726697847247124, "max_output_norm": 1.490691065788269, "near_pole_output_norm": 1.3273040056228638, "max_grad_norm": 0.0, "nan_steps": 0, "parameters": 1344 } } }