Skip to content

Conversation

@pan-x-c
Copy link
Collaborator

@pan-x-c pan-x-c commented Sep 22, 2025

Description

As the title says

Checklist

Please check the following items before code is ready to be reviewed.

  • Code has passed all tests
  • Docstrings have been added/updated in Google Style
  • Documentation has been updated
  • Code is ready for review

@gemini-code-assist
Copy link
Contributor

Summary of Changes

Hello @pan-x-c, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request introduces a robust mechanism for collecting feedback from agent applications interacting with the Explorer API. It establishes a session-based system where experiences generated during a session can be linked and later attributed with a reward, enabling more sophisticated reinforcement learning feedback loops. The changes span the API endpoints, the core service logic for managing experiences, and the client-side integration for agents.

Highlights

  • Feedback Collection API Endpoints: Introduced new API endpoints in trinity/explorer/api/api.py: /allocate for generating a unique session ID and /feedback for receiving reward feedback associated with a specific session ID. The existing chat_completions endpoint was updated to accept an optional session_id.
  • Session-based Experience Management: The ExplorerService in trinity/explorer/api/service.py was enhanced to manage experiences on a per-session basis. It now includes a session_level_experience_queue to temporarily store experiences linked to a session until feedback is received. A new allocate_session method generates session IDs, and record_feedback applies rewards to session-specific experiences before moving them to the general experience queue.
  • API Client Updates: The ExplorerClient in trinity/explorer/explorer_client.py was refactored to better handle API URLs and integrate with the new feedback mechanism. It now initializes a session upon creation and provides methods (feedback, feedback_async) to send reward feedback to the Explorer API, associating it with the client's session ID.
  • Concurrency and Type Hinting Improvements: Added an asyncio.Lock (queue_lock) to ExplorerService to ensure thread-safe access to experience queues. Several asynchronous methods in ExplorerService were updated with explicit -> None return type hints for clarity.
Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature Command Description

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

  1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution.

Copy link
Contributor

@gemini-code-assist gemini-code-assist bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

This pull request introduces a mechanism for collecting feedback from agent applications through new API endpoints. It adds session management to associate experiences with feedback. The overall implementation is sound, but I've identified a critical bug in an API call and a high-severity issue related to the lack of input validation, which could lead to server errors. My review includes suggestions to fix these issues.

@pan-x-c
Copy link
Collaborator Author

pan-x-c commented Sep 22, 2025

/unittest-all

@github-actions
Copy link

Summary

Tests 📝 Passed ✅ Failed ❌ Skipped ⏭️ Other ❓ Flaky 🍂 Duration ⏱️
142 141 0 1 0 0 2.8s

Skipped

Tests Status
tests/trainer/trainer_test.py::TestTrainerMultiModal::test_trainer skipped ⏭️

Tests

Test Name Status Flaky Duration
tests/algorithm/advantage_fn_test.py::TestGroupedAdvantageFn::test_duplicate_grpo 1ms
tests/algorithm/advantage_fn_test.py::TestGroupedAdvantageFn::test_grpo_advantage 1ms
tests/algorithm/advantage_fn_test.py::TestGroupedAdvantageFn::test_grpo_correct_bias 1ms
tests/algorithm/advantage_fn_test.py::TestGroupedAdvantageFn::test_grpo_reward_std 1ms
tests/algorithm/advantage_fn_test.py::TestGroupedAdvantageFn::test_step_wise_grpo_advantage 1ms
tests/algorithm/policy_loss_test.py::VerlPolicyLossTest::test_dpo_policy_loss 1ms
tests/algorithm/policy_loss_test.py::VerlPolicyLossTest::test_gspo_policy_loss 1ms
tests/algorithm/policy_loss_test.py::VerlPolicyLossTest::test_mix_policy_loss 1ms
tests/algorithm/policy_loss_test.py::VerlPolicyLossTest::test_opmd_policy_loss 1ms
tests/algorithm/policy_loss_test.py::VerlPolicyLossTest::test_ppo_policy_loss 1ms
tests/algorithm/policy_loss_test.py::VerlPolicyLossTest::test_sft_policy_loss 1ms
tests/buffer/experience_pipeline_test.py::TestExperiencePipeline::test_experience_pipeline 11ms
tests/buffer/experience_storage_test.py::ExperienceStorageTest::test_sql_experience_buffer 3ms
tests/buffer/experience_storage_test.py::ExperienceStorageTest::test_sql_storage_0_sft 5ms
tests/buffer/experience_storage_test.py::ExperienceStorageTest::test_sql_storage_1_dpo 5ms
tests/buffer/file_test.py::TestFileBuffer::test_file_reader 1ms
tests/buffer/file_test.py::TestFileBuffer::test_file_writer 2ms
tests/buffer/formatter_test.py::TestFormatter::test_dpo_messages_formatter 1ms
tests/buffer/formatter_test.py::TestFormatter::test_dpo_plaintext_formatter 1ms
tests/buffer/formatter_test.py::TestFormatter::test_sft_messages_formatter 1ms
tests/buffer/formatter_test.py::TestFormatter::test_sft_plaintext_formatter 1ms
tests/buffer/formatter_test.py::TestFormatter::test_task_formatter 1ms
tests/buffer/queue_test.py::TestQueueBuffer::test_priority_queue_buffer_reuse 7ms
tests/buffer/queue_test.py::TestQueueBuffer::test_priority_queue_capacity 3ms
tests/buffer/queue_test.py::TestQueueBuffer::test_queue_buffer_0_queue 4ms
tests/buffer/queue_test.py::TestQueueBuffer::test_queue_buffer_1_priority_queue 4ms
tests/buffer/queue_test.py::TestQueueBuffer::test_queue_buffer_capacity 4ms
tests/buffer/reward_shaping_mapper_test.py::TestRewardShapingMapper::test_basic_usage 1ms
tests/buffer/sql_test.py::TestSQLBuffer::test_sql_buffer_read_write 3ms
tests/buffer/task_storage_test.py::TaskStorageTest::test_read_task_0 1ms
tests/buffer/task_storage_test.py::TaskStorageTest::test_read_task_1 2ms
tests/buffer/task_storage_test.py::TaskStorageTest::test_read_task_2 1ms
tests/buffer/task_storage_test.py::TaskStorageTest::test_read_task_3 2ms
tests/buffer/task_storage_test.py::TaskStorageTest::test_read_task_4 1ms
tests/buffer/task_storage_test.py::TaskStorageTest::test_read_task_5 3ms
tests/cli/launcher_test.py::TestLauncherMain::test_main_run_command 7ms
tests/cli/launcher_test.py::TestLauncherMain::test_main_run_in_dlc 1ms
tests/cli/launcher_test.py::TestLauncherMain::test_main_studio_command 1ms
tests/cli/launcher_test.py::TestLauncherMain::test_multi_stage_run 1ms
tests/common/config_test.py::TestConfig::test_all_examples_are_valid 12ms
tests/common/config_test.py::TestConfig::test_config_flatten 1ms
tests/common/config_test.py::TestConfig::test_continue_from_checkpoint_is_valid 1ms
tests/common/config_test.py::TestConfig::test_load_default_config 4ms
tests/common/experience_test.py::TestEID::test_eid_properties 1ms
tests/common/experience_test.py::TestExperience::test_action_mask_and_logprobs_type 1ms
tests/common/experience_test.py::TestExperience::test_assertions 1ms
tests/common/experience_test.py::TestExperience::test_dpo_experience 1ms
tests/common/experience_test.py::TestExperience::test_gather 1ms
tests/common/experience_test.py::TestExperience::test_hf_datasets_conversion 1ms
tests/common/experience_test.py::TestExperience::test_multi_turn_experience 1ms
tests/common/experience_test.py::TestExperience::test_serialize_deserialize 1ms
tests/common/experience_test.py::TestExperience::test_single_turn_experience 1ms
tests/common/experience_test.py::TestExperience::test_to_dict 1ms
tests/common/experience_test.py::TestExperienceConversion::test_batch_conversion 1ms
tests/common/experience_test.py::TestExperienceConversion::test_dpo_experience_batch_conversion 1ms
tests/common/experience_test.py::TestExperienceConversion::test_experience_model_experience_conversion 1ms
tests/common/experience_test.py::TestExperienceConversion::test_gather_experiences_with_custom_fields 1ms
tests/common/experience_test.py::TestExperienceConversion::test_multiturn_experience_batch_converstion 1ms
tests/common/vllm_test.py::ModelWrapperTest_0::test_generate 55ms
tests/common/vllm_test.py::ModelWrapperTest_1::test_generate 35ms
tests/common/vllm_test.py::ModelWrapperTest_2::test_generate 45ms
tests/common/vllm_test.py::TestModelLen_0::test_model_len 21ms
tests/common/vllm_test.py::TestModelLen_1::test_model_len 20ms
tests/common/vllm_test.py::TestAPIServer::test_api 24ms
tests/common/vllm_test.py::TestAsyncAPIServer::test_api_async 24ms
tests/common/vllm_test.py::TestTokenizer::test_action_mask 1ms
tests/common/vllm_test.py::TestTokenizer::test_action_mask_with_tools 1ms
tests/common/vllm_test.py::TestAPIServerToolCall_0_deepseek_r1::test_api_tool_calls 21ms
tests/common/vllm_test.py::TestAPIServerToolCall_1::test_api_tool_calls 20ms
tests/explorer/explorer_test.py::TestExplorerCountdownEval::test_explorer 59ms
tests/explorer/explorer_test.py::TestExplorerCountdownNoEval::test_explorer 47ms
tests/explorer/explorer_test.py::TestExplorerGSM8k::test_explorer 204ms
tests/explorer/explorer_test.py::ServeTest::test_serve 66ms
tests/explorer/scheduler_test.py::SchedulerTest::test_async_workflow 4ms
tests/explorer/scheduler_test.py::SchedulerTest::test_concurrent_operations 3ms
tests/explorer/scheduler_test.py::SchedulerTest::test_get_results 22ms
tests/explorer/scheduler_test.py::SchedulerTest::test_multi_step_execution 4ms
tests/explorer/scheduler_test.py::SchedulerTest::test_non_repeatable_workflow 4ms
tests/explorer/scheduler_test.py::SchedulerTest::test_scheduler_all_methods 14ms
tests/explorer/scheduler_test.py::SchedulerTest::test_scheduler_restart_after_stop 7ms
tests/explorer/scheduler_test.py::SchedulerTest::test_split_tasks 7ms
tests/explorer/scheduler_test.py::SchedulerTest::test_stepwise_experience_eid 5ms
tests/explorer/scheduler_test.py::SchedulerTest::test_wait_all 7ms
tests/explorer/scheduler_test.py::SchedulerTest::test_wait_all_timeout_with_multi_batch 13ms
tests/explorer/step_wise_workflow_test.py::WorkflowTest::test_reward_propagation_workflow 1ms
tests/explorer/step_wise_workflow_test.py::WorkflowTest::test_step_wise_reward_workflow 1ms
tests/explorer/step_wise_workflow_test.py::WorkflowTest::test_workflows_raise_error 1ms
tests/explorer/step_wise_workflow_test.py::WorkflowTest::test_workflows_stop_at_max_env_steps 1ms
tests/explorer/workflow_test.py::WorkflowTest::test_gsm8k_workflow 1ms
tests/explorer/workflow_test.py::WorkflowTest::test_math_boxed_workflow 1ms
tests/explorer/workflow_test.py::WorkflowTest::test_math_complex_workflow 1ms
tests/explorer/workflow_test.py::WorkflowTest::test_math_eval_workflow 1ms
tests/explorer/workflow_test.py::WorkflowTest::test_math_fraction_workflow 1ms
tests/explorer/workflow_test.py::WorkflowTest::test_math_workflow 1ms
tests/explorer/workflow_test.py::WorkflowTest::test_rm_gallery_workflow 1ms
tests/explorer/workflow_test.py::WorkflowTest::test_workflow_repeatable 1ms
tests/explorer/workflow_test.py::WorkflowTest::test_workflow_resettable 1ms
tests/manager/synchronizer_test.py::TestSynchronizerExit::test_synchronizer 29ms
tests/manager/synchronizer_test.py::TestStateDictBasedSynchronizer_0::test_synchronizer 75ms
tests/manager/synchronizer_test.py::TestStateDictBasedSynchronizer_1::test_synchronizer 72ms
tests/manager/synchronizer_test.py::TestStateDictBasedSynchronizer_2::test_synchronizer 116ms
tests/manager/synchronizer_test.py::TestStateDictBasedSynchronizer_3::test_synchronizer 132ms
tests/manager/synchronizer_test.py::TestNCCLBasedSynchronizer_0::test_synchronizer 70ms
tests/manager/synchronizer_test.py::TestNCCLBasedSynchronizer_1::test_synchronizer 70ms
tests/service/data_juicer_test.py::TestDataJuicer::test_config 1ms
tests/service/data_juicer_test.py::TestDataJuicer::test_server_start 22ms
tests/service/data_juicer_test.py::TestDataJuicerExperiencePipeline::test_data_juicer_operators 21ms
tests/service/data_juicer_test.py::TestDataJuicerTaskPipeline::test_data_juicer_task_pipeline 14ms
tests/trainer/trainer_test.py::TestTrainerCountdown_0_fsdp::test_trainer 153ms
tests/trainer/trainer_test.py::TestTrainerCountdown_1_megatron::test_trainer 317ms
tests/trainer/trainer_test.py::TestStepAheadAsyncRL::test_trainer 58ms
tests/trainer/trainer_test.py::TestTrainerGSM8K_0_fsdp::test_trainer 52ms
tests/trainer/trainer_test.py::TestTrainerGSM8K_1_fsdp2::test_trainer 55ms
tests/trainer/trainer_test.py::TestTrainerGSM8K_2_fsdp::test_trainer 57ms
tests/trainer/trainer_test.py::TestTrainerGSM8K_3_fsdp2::test_trainer 61ms
tests/trainer/trainer_test.py::TestTrainerSFTWarmupGSM8K::test_trainer 102ms
tests/trainer/trainer_test.py::TestTrainerDPO::test_trainer 39ms
tests/trainer/trainer_test.py::TestTrainerSFT::test_trainer 36ms
tests/trainer/trainer_test.py::TestTrainerToolsSFT::test_trainer_tools 36ms
tests/trainer/trainer_test.py::TestFullyAsyncMode_0_fsdp::test_fully_async_mode 81ms
tests/trainer/trainer_test.py::TestFullyAsyncMode_1_fsdp::test_fully_async_mode 77ms
tests/trainer/trainer_test.py::TestFullyAsyncMode_2_megatron::test_fully_async_mode 171ms
tests/trainer/trainer_test.py::TestTrainerMIX::test_trainer 56ms
tests/trainer/trainer_test.py::TestTrainerMultiModal::test_trainer ⏭️ 1ms
tests/utils/eval_utils_test.py::TestComputeScore::test_both_boxed_and_equivalent 1ms
tests/utils/eval_utils_test.py::TestComputeScore::test_both_boxed_and_not_equivalent 1ms
tests/utils/eval_utils_test.py::TestComputeScore::test_empty_ground_truth 1ms
tests/utils/eval_utils_test.py::TestComputeScore::test_empty_solution_string 1ms
tests/utils/eval_utils_test.py::TestComputeScore::test_multiple_boxed_answers_in_solution 1ms
tests/utils/eval_utils_test.py::TestComputeScore::test_solution_boxed_truth_raw_and_equivalent 1ms
tests/utils/eval_utils_test.py::TestComputeScore::test_solution_boxed_truth_raw_and_not_equivalent 1ms
tests/utils/eval_utils_test.py::TestComputeScore::test_solution_not_boxed 1ms
tests/utils/eval_utils_test.py::TestComputeScore::test_solution_raw_and_ground_truth_boxed_equivalent 1ms
tests/utils/eval_utils_test.py::TestMathEvalUtils::test_extract_answer 1ms
tests/utils/eval_utils_test.py::TestMathEvalUtils::test_verify_math_answer 1ms
tests/utils/eval_utils_test.py::TestEvalUtils::test_is_equiv 1ms
tests/utils/log_test.py::LogTest::test_actor_log 2ms
tests/utils/log_test.py::LogTest::test_group_by_node 2ms
tests/utils/log_test.py::LogTest::test_no_actor_log 1ms
tests/utils/plugin_test.py::TestPluginLoader::test_load_plugins_local 1ms
tests/utils/plugin_test.py::TestPluginLoader::test_load_plugins_remote 6ms
tests/utils/plugin_test.py::TestPluginLoader::test_passing_custom_class 3ms

Github Test Reporter by CTRF 💚

@pan-x-c pan-x-c changed the title [WIP] Explorer API collects feedback from agent applications Explorer API collects feedback from agent applications Oct 15, 2025
@pan-x-c
Copy link
Collaborator Author

pan-x-c commented Oct 15, 2025

/unittest-all

@pan-x-c
Copy link
Collaborator Author

pan-x-c commented Oct 21, 2025

/unittest-all

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant