1 Pull request merged by 1 person

• add eval scripts to reproduce results of Qwen3

  #1565 merged Jul 23, 2025

1 Pull request opened by 1 person

• Refactor: Improve modularity and error handling in eval.py

  #1570 opened Jul 24, 2025

9 Issues closed by 4 people

• [Badcase]: Qwen3-30B-A3B 在 bfcl v3 中multi_turn base 子集的复现分数与 leaderboard 不一致，能否告知测试时使用的推理参数

  #1518 closed Jul 24, 2025

• [Bug]: 网页版qwen3会将$$识别成\[或\]或\(或\)

  #1431 closed Jul 22, 2025

• When Qwen-3-32-Base on huggingface?

  #1546 closed Jul 22, 2025

• [Badcase]: Hallucination on Tengwang Ge Xu - 背写滕王阁序

  #1448 closed Jul 22, 2025

• [Badcase]: Some case no thinking output,

  #1491 closed Jul 22, 2025

• qwen什么时候可以支持上下文长度128K

  #1558 closed Jul 21, 2025

• 可以公开qwen3评估多语言时所使用的fewshot和提示吗？

  #1494 closed Jul 20, 2025

• [Bug]: Loading the Qwen3-30B-A3B-GPTQ-Int4 model using Transformers is very slow, and it cannot perform inference properly

  #1473 closed Jul 19, 2025

• 重复解码问题

  #1476 closed Jul 19, 2025

11 Issues opened by 11 people

• Number of samples used to estimate pass@1 scores?

  #1571 opened Jul 24, 2025

• [REQUEST]: 能否将模型转换为FLAX类型以提升性能，有没相关的实现。

  #1569 opened Jul 24, 2025

• vllm启动QwQ以及Qwen3-32B 模型，使用functioncall功能时，流式调用接口时，argument变量返回始终只能得到一个值，而非流式调用时一切正常

  #1568 opened Jul 24, 2025

• [Badcase]: Performance drop using rope scaling with Qwen3-8b in vllm

  #1567 opened Jul 24, 2025

• How to check which expert is activated at each layer during Qweb 1.5 MoE inference

  #1566 opened Jul 23, 2025

• Login/account creation problems with qwen chat

  #1564 opened Jul 23, 2025

• Low IFEval in Qwen3

  #1563 opened Jul 23, 2025

• Release of Instruction-tuned (IT) version of Qwen3 for difference sizes

  #1562 opened Jul 22, 2025

• Bug: Incorrect spacing before punctuation in Russian language output

  #1561 opened Jul 22, 2025

• 请问是用什么评分模型测 AlignBench 的？

  #1557 opened Jul 21, 2025

• [REQUEST]: Qwen3 Knowledge Cutoff Date

  #1554 opened Jul 18, 2025

35 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• [Bug]: Qwen3-32B有时候思考过程只有开头的<think>，结尾没有</think>

  #1459 commented on Jul 18, 2025 • 0 new comments

• [Badcase]: Infinite Repetition When Listing Emojis

  #1553 commented on Jul 21, 2025 • 0 new comments

• [Badcase]: Thinking language not follow user chat language

  #1550 commented on Jul 21, 2025 • 0 new comments

• Qwen3训练的时候如果不要带思考能力，数据集是否可以不带/think标签，如果数据集全部不带/think标签是否会影响模型原本的能力

  #1487 commented on Jul 22, 2025 • 0 new comments

• [Request ] 32B (Dense) Base Model?

  #1275 commented on Jul 22, 2025 • 0 new comments

• [Badcase]: Can anyone reproduce the effect of the qwen3-Base model on GSM8K?

  #1540 commented on Jul 23, 2025 • 0 new comments

• [REQUEST]: 你好，想问一下技术报告中qwen3 base测试结果，使用了的解码参数是什么样的，是使用的贪心解码吗？

  #1468 commented on Jul 23, 2025 • 0 new comments

• [REQUEST]: self-study

  #1497 commented on Jul 23, 2025 • 0 new comments

• [REQUEST]: reading files

  #1498 commented on Jul 23, 2025 • 0 new comments

• [REQUEST]: the creator of the programs

  #1504 commented on Jul 23, 2025 • 0 new comments

• [REQUEST]: The memory effect

  #1507 commented on Jul 23, 2025 • 0 new comments

• [REQUEST]: working on the speed of responses

  #1513 commented on Jul 23, 2025 • 0 new comments

• [REQUEST]: increase the number of uploaded files to 10

  #1517 commented on Jul 23, 2025 • 0 new comments

• [REQUEST]: The length of the tokens.

  #1527 commented on Jul 23, 2025 • 0 new comments

• [REQUEST]: deleting chats

  #1528 commented on Jul 23, 2025 • 0 new comments

• [REQUEST]: teach qwen app development

  #1532 commented on Jul 23, 2025 • 0 new comments

• [REQUEST]: teach the model to fix CVE

  #1534 commented on Jul 23, 2025 • 0 new comments

• request

  #1535 commented on Jul 23, 2025 • 0 new comments

• [REQUEST]: focus on programming

  #1542 commented on Jul 23, 2025 • 0 new comments

• [REQUEST]: smoothness of the interface

  #1545 commented on Jul 23, 2025 • 0 new comments

• [REQUEST]: follow links

  #1551 commented on Jul 23, 2025 • 0 new comments

• [Badcase]: context inject attack

  #1547 commented on Jul 23, 2025 • 0 new comments

• [REQUEST]: Qwen Chat - improved Web Search tool

  #1524 commented on Jul 23, 2025 • 0 new comments

• [QwenChat] Qwen unreasonably blocking query and locking out future use

  #1475 commented on Jul 23, 2025 • 0 new comments

• [QwenChat]: can't create account

  #1306 commented on Jul 23, 2025 • 0 new comments

• [Bug]: 从书里直接复制阿拉伯语这种RTL的语言翻译错误

  #1549 commented on Jul 23, 2025 • 0 new comments

• [Badcase]: Poor quality of outputs in Polish language

  #919 commented on Jul 23, 2025 • 0 new comments

• [Badcase]: Poor quality of outputs in Russian language

  #928 commented on Jul 23, 2025 • 0 new comments

• [Badcase]: Qwen2.5-7B-Instruct 中德、中意翻译 不遵循指令、code switch

  #1097 commented on Jul 23, 2025 • 0 new comments

• [Bug]: Poor support for Hebrew

  #1114 commented on Jul 23, 2025 • 0 new comments

• [Badcase]: 进行中文-英文翻译出现问题

  #1253 commented on Jul 23, 2025 • 0 new comments

• [Badcase]: Fail to reproduce the results of Qwen3-32B on GPQA diamond.

  #1503 commented on Jul 23, 2025 • 0 new comments

• [Badcase]: qwen3 平台适配问题

  #1516 commented on Jul 24, 2025 • 0 new comments

• [Badcase]: Qwen3-32B基准测评

  #1501 commented on Jul 24, 2025 • 0 new comments

• Qwen3-4B 性能和tech report对不上

  #1483 commented on Jul 24, 2025 • 0 new comments