grpo

Grouped Reward Policy Optimization, a reinforcement learning algorithm for training language models with group-based reward estimation.

Name	Categories	Topics	References	Credibility	Title	RelatedNotes	NavigationOrder	Eleventy	TemplateEngineOverride
20260509_skillos_learning_skill_curation_self_evolving_agents	paper_analysis	agent_architecture, memory_mechanism, reinforce_learning, self_evolving_agents, skill_curation	google_cloud_ai_research, grpo, skillos, alfworld	100	SkillOS: Learning Skill Curation for Self-Evolving Agents	self_evolving_agents_survey_asi, memory_os_of_ai_agent, agentevolver_self_evolving_agent, reasoningbank_scaling_agent_self_evolving_reasoning_memory	[object Object]	[object Object]	md
20260520_amr_sd_token_level_credit_assignment	paper_analysis	reinforce_learning, reasoning, llm, reward_modeling, self_evolving_agents	grpo, dapo, meituan, sciknoweval	100	AMR-SD: Asymmetric Meta-Reflective Self-Distillation for Token-Level Credit Assignment	rl_long_horizon_reasoning_llm_expressiveness, skillos_learning_skill_curation_self_evolving_agents	[object Object]	[object Object]	md
20260524_vector_policy_optimization_vpo	paper_analysis	reinforce_learning, llm, test_time_scaling, reasoning, reward_modeling	mit, grpo	100	Vector Policy Optimization: Training for Diversity Improves Test-Time Search		[object Object]	[object Object]	md
20260528_CORE_contrastive_reflection_reasoning	paper_analysis	reasoning, memory_mechanism, self_evolving_agents, contrastive_reflection, reinforce_learning, cognitive_science	stanford_iris_lab, grpo, memgpt	100	CORE: Contrastive Reflection Enables Rapid Improvements in Reasoning	Can RL Teach Long-Horizon Reasoning to LLMs? Expressiveness Is Key, MemLineage: Lineage-Guided Enforcement for LLM Agent Memory, AMR-SD: Asymmetric Meta-Reflective Self-Distillation for Token-Level Credit Assignment	[object Object]	[object Object]	md
20260609_socratic_swe_self_evolving_coding_agents_trace_derived_skills	paper_analysis	self_evolving_agents, memory_mechanism, skill_curation, reinforce_learning, reasoning	sjtu, swebench_verified, terminalbench_2, grpo, skillos	100	Socratic-SWE: Self-Evolving Coding Agents via Trace-Derived Agent Skills	skillos_learning_skill_curation_self_evolving_agents, skillopt_executive_strategy_self_evolving_agent_skills, aflow_automating_agentic_workflow_generation, scaling_harness_agentic_ai, meme_multi_entity_evolving_memory_evaluation	[object Object]	[object Object]	md
20260627_joint_learning_experiential_rules_policies_llm_agents	paper_analysis	memory_mechanism, reinforce_learning, self_evolving_agents, reasoning, agent_architecture	alfworld, grpo, sun_yat_sen_university	100	Joint Learning of Experiential Rules and Policies for Large Language Model Agents	unlocking_working_memory_latent_reasoning, memory_os_of_ai_agent, memskill_learning_evolving_memory_skills	[object Object]	[object Object]	md
20260629_memory_r1_enhancing_llm_agents_manage_utilize_memories_rl	paper_analysis	agent_architecture, memory_mechanism, reinforce_learning, llm, long_term_memory	lmu_munich, mem0, locomo_bench, grpo	100	Memory-R1: Enhancing Large Language Model Agents to Manage and Utilize Memories via Reinforcement Learning	memgpt_towards_llms_as_operating_systems, muse_autoskill_self_evolving_skill_memory, gems_agent_native_multimodal_generation_memory_skills, amr_sd_token_level_credit_assignment, agent_memory_characterization_system_implications	[object Object]	[object Object]	md
20260701_severa_verified_synthesis_self_evolving_agents	paper_analysis	agent_architecture, self_evolving_agents, reinforce_learning, reasoning, agent_security	university_of_illinois_urbana_champaign, grpo, tau_squared_bench	100	SEVerA: Verified Synthesis of Self-Evolving Agents	skillos_learning_skill_curation_self_evolving_agents, muse_autoskill_self_evolving_skill_memory, trace_unified_rollout_budget_agentic_rl, amr_sd_token_level_credit_assignment, agent_memory_characterization_system_implications	[object Object]	[object Object]	md
20260706_demopsd_disagreement_modulated_policy_self_distillation	paper_analysis	reinforce_learning, reasoning, llm, self_evolving_agents	grpo, sciknoweval, kl_distillation, gpqa	100	DemoPSD: Disagreement-Modulated Policy Self-Distillation	rl_long_horizon_reasoning, vector_policy_optimization	[object Object]	[object Object]	md