MARTI
A Framework for LLM-based Multi-Agent Reinforced Training and Inference
파일 탐색기
최종 버전 다운로드 (.zip)- arxiv.png
- ds-1.5-qwen-1.7-avg.jpg
- framework.jpg
- heter_pass1.png
- homo_results.png
- logo.jpg
- mad-dynamics.jpg
- mad-rl-amc.jpg
- mad-rl-math.jpg
- mars2_framework.png
- marti-multi-turn-agent.jpg
- marti-workflow-config.jpg
- moa-dynamics.jpg
- moa-rl-amc.jpg
- moa-rl-math.jpg
- qwen2.5-3b-base-instruct-avg.jpg
- workflow.jpg
- test.jsonl
- train.jsonl
- test.jsonl
- train.jsonl
- aime.json
- amc.json
- gpqa.json
- math.json
- minerva.json
- olympiad_bench.json
- Qwen2.5-3B-Instruct_MATH_coa@3_results.json
- Qwen2.5-3B-Instruct_MATH_mad@4_results.json
- Qwen2.5-3B-Instruct_MATH_mix@4_results.json
- test.json
- train.json
- test.json
- train.json
- 1-Overview-Of-MARTI.md
- 2-Workflows-Integration.md
- 3-Reward-And-Training.md
- 4-Experiments-Of-MARTI.md
- multi_agent_ppotrainer.py
- multi_agent_samples_generator.py
- multi_agent_train_ppo_ray.py
- run_train_multi_mcts.sh
- run_train_single_mcts.sh
- run_train_chain.sh
- run_train_mad.sh
- run_train_mathchat.sh
- run_train_search_r1.sh
- run_train_tir.sh
- agent_func.py
- agent_func_code.py
- agent_func_gem_multiturn.py
- reward_func.py
- run_train_reviewrl_async.sh
- calculator.json
- code.json
- google.json
- python.json
- search.json
- run_train_code_async.sh
- run_train_math_async.sh
- base_step.py
- code_step.py
- mcp_step.py
- AsyncABMCTSA.py
- base.py
- prompt_configs.py
- task.py
- task.py
- base.py
- __init__.py
- data_types.py
- eval_result.py
- llm_generation_interface.py
- model_base.py
- utils.py
- proc_lcb_results.py
- visualize.py
- code_prompt.py
- math_prompt.py
- utils.py
- __init__.py
- base.py
- chain.py
- debate.py
- mixture.py
- openai_api.py
- prover_verifier.py
- reviewrl_genrm.py
- sandbox_fusion.py
- search_r1.py
- base.py
- calculator.py
- judge.py
- manager.py
- mcp_manager.py
- parser.py
- sandbox.py
- search.py
- utils.py
- ab_mcts_processor.py
- ab_mcts_workflow.py
- chain_processor.py
- chain_workflow.py
- debate_processor.py
- debate_workflow.py
- default_processor.py
- default_workflow.py
- experience_with_tool_workflow.py
- experience_with_tool_workflow1.py
- experience_workflow.py
- judge_workflow.py
- mathchat_workflow.py
- single_codeworkflow.py
- single_mathworkflow.py
- tool_workflow.py
- utils.py
- workflow_wrapper.py
- __init__.py
- batch_inference.py
- interactive_chat.py
- lora_combiner.py
- multi_agent_train_ppo_ray.py
- serve_rm.py
- train_dpo.py
- train_kd.py
- train_kto.py
- train_ppo_ray.py
- train_prm.py
- train_rm.py
- train_sft.py
- __init__.py
- process_reward_dataset.py
- prompts_dataset.py
- reward_dataset.py
- sft_dataset.py
- unpaired_preference_dataset.py
- utils.py
- __init__.py
- actor.py
- loss.py
- model.py
- ring_attn_utils.py
- utils.py
- __init__.py
- experience_maker.py
- experience_maker_async.py
- kl_controller.py
- multi_agent_experience_maker.py
- multi_agent_samples_generator.py
- replay_buffer.py
- __init__.py
- launcher.py
- ppo_actor.py
- ppo_critic.py
- utils.py
- vllm_engine.py
- vllm_engine_async.py
- vllm_worker_wrap.py
- __init__.py
- dpo_trainer.py
- kd_trainer.py
- kto_trainer.py
- multi_agent_ppotrainer.py
- ppo_trainer.py
- ppo_trainer_async.py
- prm_trainer.py
- rm_trainer.py
- sft_trainer.py
- __init__.py
- deepspeed.py
- deepspeed_utils.py
- __init__.py
- agent.py
- distributed_sampler.py
- distributed_util.py
- logging_utils.py
- processor.py
- remote_rm_utils.py
- seqlen_balancing.py
- utils.py
- testing_util.py
- __init__.py
- code_reward.py
- __init__.py
- codeforces.py
- codeforces_test.py
- firejail_exec.py
- humanevalplus.py
- kodcode.py
- livecodebench.py
- pyext2.py
- taco.py
- utils.py
- __init__.py
- code_reward.py
- reward_types.py
- __init__.py
- utils.py
- __init__.py
- globals.py
- math_reward.py
- reward_types.py
- main.py
- __init__.py
- grader.py
- math_grade.py
- math_normalize.py
- qwen_eval.py
- qwen_eval_timeout.py
- qwen_math_parser.py
- simplerl_math_equal.py
- review_eval.py
- review_eval.py
- qa_em.py
- qa_em_boxed.py
- qa_em_format.py
- __init__.py
- auto_reward_alloc.py
- auto_verify.py
- __init__.py
- .gitignore
- LICENSE
- README.md
- requirements.txt
- setup.py
- version.txt
// repository documentation
Was this content helpful?
(0 ratings)