Issues

• math_equal accepts numeric predictions that are 100x or 1/100 of the reference, and the evaluation script scores with it

  hkust-nlp/simpleRL-reason#101 · Filter by author shaurya416 opened on Sep 25, 2026

• Performance of Qwen-2.5-32B-SimpleRL-Zoo

  hkust-nlp/simpleRL-reason#99 · Filter by author sjavaheri opened on Feb 25, 2026

• Actor was died.

  hkust-nlp/simpleRL-reason#98 · Filter by author dct-cell opened on Dec 10, 2025
  1

• On which benchmark did the results of plot_figure1_v2.3_token_length_vs_steps.png come from?

  hkust-nlp/simpleRL-reason#97 · Filter by author dzh19990407 opened on Nov 27, 2025
  1

• [Feedback] Replication Report: Solved Dependency Hell, Data Loading, and Hardcoding Issues

  hkust-nlp/simpleRL-reason#96 · Filter by author HappynessI opened on Nov 16, 2025
  4

• Timeout error of math_verify

  hkust-nlp/simpleRL-reason#95 · Filter by author yuxiaooye opened on Nov 14, 2025

• Ray runtime error fix: install specific Click version

  hkust-nlp/simpleRL-reason#94 · Filter by author XuGW-Kevin opened on Nov 9, 2025

• evaluation script to test Meta-Llama-3.1 8B(base model step 0) on GSM8K,results differ significantly acc 39.7(your) vs 24.1(mine)

  hkust-nlp/simpleRL-reason#92 · Filter by author AILuckyStar opened on Aug 3, 2025
  2

• Question: Response length not increase

  hkust-nlp/simpleRL-reason#90 · Filter by author stonewst opened on Jul 28, 2025
  1

• Evaluation Details on IFEVAL, MMLU, and GPQA-Diamond

  hkust-nlp/simpleRL-reason#88 · Filter by author zijunmin opened on Jul 3, 2025
  1

• plan to support NPU ？

  hkust-nlp/simpleRL-reason#82 · Filter by author wandouguo opened on Apr 27, 2025
  1

• Actor died without any reason

  hkust-nlp/simpleRL-reason#80 · Filter by author shuminwang-ai opened on Apr 16, 2025
  5

• math_verify and latex2sympy2_extended need python>=3.10,

  hkust-nlp/simpleRL-reason#77 · Filter by author zyf8818 opened on Apr 15, 2025
  4

• Why does using Qwen-2.5-1.5B generate garbled text during training and produce particularly poor results?

  hkust-nlp/simpleRL-reason#67 · Filter by author LLMforScience opened on Apr 2, 2025
  13

• Question about Response Length

  hkust-nlp/simpleRL-reason#65 · Filter by author ayyyq opened on Apr 1, 2025
  4

• Question about AIME24 Evaluation

  hkust-nlp/simpleRL-reason#64 · Filter by author Zhuofeng-Li opened on Apr 1, 2025
  1

• Did you guys use a cloud provider?

  hkust-nlp/simpleRL-reason#63 · Filter by author aidando73 opened on Apr 1, 2025
  5

• can I have the training config for Qwen 0.5B and 1.5B

  hkust-nlp/simpleRL-reason#61 · Filter by author vengdeng opened on Apr 1, 2025
  14

• model context length should be greater than total sequence length

  hkust-nlp/simpleRL-reason#58 · Filter by author Su-my opened on Mar 30, 2025
  2

• Evaluation Results

  hkust-nlp/simpleRL-reason#52 · Filter by author Wloner0809 opened on Mar 27, 2025
  2

• Intermediate checkpoints?

  hkust-nlp/simpleRL-reason#47 · Filter by author sz85512678 opened on Mar 18, 2025
  2

• Questions about evaluation results

  hkust-nlp/simpleRL-reason#46 · Filter by author yingyibiao opened on Mar 7, 2025
  5

• Eval code for the hkust-nlp/Qwen-2.5-Math-7B-SimpleRL-Zero

  hkust-nlp/simpleRL-reason#45 · Filter by author yifan1130 opened on Feb 28, 2025
  1

• Is the whole MATH training set corresponds to around 7 steps?

  hkust-nlp/simpleRL-reason#41 · Filter by author HCHCXY opened on Feb 20, 2025
  2

• Use reinforce

  hkust-nlp/simpleRL-reason#35 · Filter by author supermancmk opened on Feb 15, 2025
  2