Issues

• Validate Observation representation invariant at construction time

  AgentR1/Agent-R1#108 · Filter by author Andrew9603 opened on Sep 15, 2026

• 有Agent_Tool的训练结果吗

  AgentR1/Agent-R1#104 · Filter by author jzl20 opened on Aug 1, 2026

• Model Saving

  AgentR1/Agent-R1#102 · Filter by author whyseu opened on Jul 17, 2026
  2

• Error encountered while training on HotpotQA data

  AgentR1/Agent-R1#101 · Filter by author whyseu opened on Jul 16, 2026
  1

• Agent-R1 is compatible with 25,000+ tools via Nerq Gateway

  AgentR1/Agent-R1#74 · Filter by author agentindex-bot opened on Mar 15, 2026

• subem or exact em

  AgentR1/Agent-R1#71 · Filter by author williamd4112 opened on Jan 12, 2026

• discord link from Readme does not work

  AgentR1/Agent-R1#70 · Filter by author ArturTanona opened on Nov 21, 2025

• 这个项目还在维护么？

  AgentR1/Agent-R1#69 · Filter by author xichengpro opened on Nov 14, 2025
  1

• 关于 QuickStart run_ppo_hotpotqa.sh 报错

  AgentR1/Agent-R1#68 · Filter by author TDY-raedae opened on Sep 10, 2025

• Dependency conflicts: Verl requires tensordict<=0.6.2 but Torch 2.7.0 (CUDA 12.8, B200 GPUs) requires tensordict>=0.7.0

  AgentR1/Agent-R1#67 · Filter by author jatinS-dev opened on Sep 6, 2025
  1

• 有训练的基线结果吗

  AgentR1/Agent-R1#66 · Filter by author yangdongdong2000 opened on Aug 21, 2025
  1

• compute_grpo_outcome_advantage 的疑问

  AgentR1/Agent-R1#65 · Filter by author gyy8426 opened on Aug 18, 2025

• 如何适配llama系列模型？

  AgentR1/Agent-R1#64 · Filter by author xjtupy opened on Aug 11, 2025
  3

• ValueError: operands could not be broadcast together with shapes (243,) (158,)

  AgentR1/Agent-R1#63 · Filter by author zwxandy opened on Aug 7, 2025
  1

• How to use multiple tools during RL training?

  AgentR1/Agent-R1#62 · Filter by author zwxandy opened on Aug 4, 2025
  1

• 任务适配输出异常

  AgentR1/Agent-R1#61 · Filter by author xjtupy opened on Aug 1, 2025
  16

• 关于stop_token_ids和stop的问题

  AgentR1/Agent-R1#60 · Filter by author xjtupy opened on Jul 24, 2025
  2

• 训练显存占用

  AgentR1/Agent-R1#57 · Filter by author ZepengDu opened on Jun 30, 2025
  2

• env_rewards在代码中没有看到相关的逻辑

  AgentR1/Agent-R1#56 · Filter by author Chen-Song opened on Jun 25, 2025

• The QR code of the wechat group is expired 6-12

  AgentR1/Agent-R1#55 · Filter by author Changheng-hild opened on Jun 12, 2025
  1

• max_prompt_length，max_response_length，max_response_length_single_turn和tool.max_tool_response_length这几个参数分别表示什么作用呢，我在训练时应该如何设置这几个参数

  AgentR1/Agent-R1#54 · Filter by author myjpgit opened on Jun 9, 2025
  3

• 在上下文传递过程中，tool_response中搜索到的外部知识会随着轮次一直向下传递吗？是否会引起上下文超长

  AgentR1/Agent-R1#53 · Filter by author myjpgit opened on Jun 9, 2025
  4

• I am getting aime2024 result lower after ppo train

  AgentR1/Agent-R1#52 · Filter by author lebronjamesking opened on Jun 5, 2025

• what is the difference between your two branch debug-nan and main?

  AgentR1/Agent-R1#50 · Filter by author 141forever opened on May 30, 2025

• I didn't put the tool in the training call, why does the model after Qwen2-14b-instruct training have the fields <translation> and </translation> when reasoning? Is it because the Qwen pre-training process adds a lot of similar data training?

  AgentR1/Agent-R1#49 · Filter by author shaoxiong007 opened on May 30, 2025