{"ckpt": "checkpoint-200", "sets": {"vuln_detection_test": {"acc": 0.6, "base": 0.66, "delta": -0.06, "unp": 1, "f1": 0.345, "f1_base": 0.556}, "knowledge_mcq": {"acc": 0.94, "base": 0.96, "delta": -0.02, "unp": 0}, "mmlu_security": {"acc": 0.92, "base": 0.86, "delta": 0.06, "unp": 0}, "mmlu_general": {"acc": 0.78, "base": 0.66, "delta": 0.12, "unp": 4}}} {"ckpt": "checkpoint-400", "sets": {"vuln_detection_test": {"acc": 0.52, "base": 0.66, "delta": -0.14, "unp": 1, "f1": 0.667, "f1_base": 0.556}, "knowledge_mcq": {"acc": 1.0, "base": 0.96, "delta": 0.04, "unp": 0}, "mmlu_security": {"acc": 0.86, "base": 0.86, "delta": 0.0, "unp": 0}, "mmlu_general": {"acc": 0.8, "base": 0.66, "delta": 0.14, "unp": 0}}} {"ckpt": "checkpoint-600", "sets": {"vuln_detection_test": {"acc": 0.56, "base": 0.66, "delta": -0.1, "unp": 1, "f1": 0.656, "f1_base": 0.556}, "knowledge_mcq": {"acc": 1.0, "base": 0.96, "delta": 0.04, "unp": 0}, "mmlu_security": {"acc": 0.86, "base": 0.86, "delta": 0.0, "unp": 0}, "mmlu_general": {"acc": 0.88, "base": 0.66, "delta": 0.22, "unp": 1}}} {"ckpt": "checkpoint-800", "sets": {"vuln_detection_test": {"acc": 0.58, "base": 0.66, "delta": -0.08, "unp": 1, "f1": 0.655, "f1_base": 0.556}, "knowledge_mcq": {"acc": 0.98, "base": 0.96, "delta": 0.02, "unp": 0}, "mmlu_security": {"acc": 0.86, "base": 0.86, "delta": 0.0, "unp": 0}, "mmlu_general": {"acc": 0.76, "base": 0.66, "delta": 0.1, "unp": 4}}} {"ckpt": "checkpoint-1000", "sets": {"vuln_detection_test": {"acc": 0.62, "base": 0.66, "delta": -0.04, "unp": 1, "f1": 0.679, "f1_base": 0.556}, "knowledge_mcq": {"acc": 1.0, "base": 0.96, "delta": 0.04, "unp": 0}, "mmlu_security": {"acc": 0.86, "base": 0.86, "delta": 0.0, "unp": 0}, "mmlu_general": {"acc": 0.76, "base": 0.66, "delta": 0.1, "unp": 6}}}