qy1026

UnboundLocalError: local variable 'prompt_token_num' referenced before assignment and NO OUTPUTS

2

### 先决条件 - [X] 我已经搜索过 [问题](https://github.com/open-compass/opencompass/issues/) 和 [讨论](https://github.com/open-compass/opencompass/discussions) 但未得到预期的帮助。 - [X] 错误在 [最新版本](https://github.com/open-compass/opencompass) 中尚未被修复。 ### 问题类型 我正在使用官方支持的任务/模型/数据集进行评估。 ### 环境 ``` {'CUDA available': True, 'CUDA_HOME': '/usr/local/cuda-12.2', 'GCC': 'gcc (Ubuntu 9.4.0-1ubuntu1~20.04.3) 9.4.0',...

CMB + Qwen1.5-72B-Chat got empty answers

1

### Prerequisite - [X] I have searched [Issues](https://github.com/open-compass/opencompass/issues/) and [Discussions](https://github.com/open-compass/opencompass/discussions) but cannot get the expected help. - [X] The bug has not been fixed in the [latest version](https://github.com/open-compass/opencompass). ### Type...

fsdp + DPO + fullyfintune会报错

3

### Reminder - [X] I have read the README and searched the existing issues. ### System Info pass ### Reproduction ``` CUDA_VISIBLE_DEVICES="0,1,2,3,4,5,6,7" accelerate launch \ --config_file examples/accelerate/fsdp_config.yaml \ src/train.py examples/train_full/llama3_full_dpo_fsdp.yaml...

``` register_template( template_name='llama2', system_format='\n{content}\n\n\n', user_format='[INST]{content}[/INST]', assistant_format='{content} ', system="You are a helpful, respectful and honest assistant. " "Always answer as helpfully as possible, while being safe. " "Your answers should not...

Is it normal the pipeline start with a huge loss ?

3

``` step 10: {'loss': 119743.8516, 'grad_norm': 938286.7284407256, 'learning_rate': 2.0161290322580643e-09, 'rewards/chosen': 0.0, 'rewards/rejected': 0.0, 'rewards/accuracies': 0.0, 'rewards/margins': 0.0, 'logps/rejected': -128.30323791503906, 'logps/chosen': -178.66146850585938, 'logits/rejected': -0.7681801915168762, 'logits/chosen': -0.792536735534668, 'epoch': 0.0} step 20: {'loss':...

When use `accelerate==0.23.0` in `setup.py`, it got the following error: `Accelerator.__init__() got an unexpected keyword argument use_seedable_sampler` When upgrade `accelerate` to 0.31.0, this error fixed.

NotImplementedError('load from hdfs is not implemented yet')

1

Change code in `verl/trainer/ppo/ray_trainer.py` from `NotImplementedError('load from hdfs is not implemented yet') ` to `raise NotImplementedError('load from hdfs is not implemented yet') `？ 否则报错内容和出错的地方不匹配。

RuntimeError: The size of tensor a (128) must match the size of tensor b (1698) at non-singleton dimension 4

4

My train script: ```shell deepspeed --include localhost:4,5,6,7 openrlhf/cli/train_prm.py \ --save_path $save_path \ --save_steps 500 \ --logging_steps 1 \ --eval_steps 5 \ --train_batch_size 8 \ --micro_train_batch_size 2 \ --pretrain /xxxx/Qwen2.5-7B-Instruct/ \...