Repository Issue Activity (beta)

#402 Can the current black-box GAD addition be directly used in Qwen 3.5?

#418 MiniLLM : Results on test split between evaluate() in finetune.py and evaluate_main() in evaluate_main.py are different

#287 assert all((~torch.isinf(scores.view(-1))) & (~torch.isnan(scores.view(-1)))) [rank0]: AssertionError

#332 [MiniLLM] The download link for OpenWebText is invalid.

#354 【GAD】使用Qwen3-VL进行GAD训练极慢

#353 [GAD] Question regarding Reward Score fluctuations and downstream evaluation

#351 Please release the test data used in the paper?

#352 [GAD] Plans to release the test data?

#355 Puzzle about the results.

#312 Can Data Selection via Optimal Control be applied in instance segmentation?

#329 [MiniLLM] Strange Rouge-L result on Dolly dataset

#333 [MiniLLM] baseline hyperparameter settings

#349 GAD在VLM的适配

#347 [GAD] Question about warmup

#344 [GAD] Unable to setup local environment with given spec

#346 [GAD] Question about SeqKD implementation in the repo

#304 [ProTeGi]where 's the file ethos_ishate_binary_shuf.csv?

#342 [GAD] verl local installation (latest) vs docker image

#330 No module named 'minillm'

#176 ImportError: cannot import name 'mpu' from 'transformers'

#328 [MiniLLM] Why the teacher prob are not extract offline?

#327 the wrong of the ProTeGi

#326 我尝试用论文《DATA SELECTION VIA OPTIMAL CONTROL FOR LANGUAGE MODELS》方法去筛选带有错误标签的数据，发现错误得标签也有可能会有较高得得分。

#325 ProTeGi：When using UCB evaluation, how many eval_rounds are set?

#323 [ProTeGi] python environment