Zhang Cong comments

Results 9 comments of


                                            Zhang Cong

使用您的代码，我在lfw的得分可以达到99.6，但是实际应用场景会出现大批量的认错

> lfw的结果高并说明不了什么，能在megaface上刷高分数才行，还有就是要看你使用的什么训练集请问adgdb-30 cfp 测试集在哪下载呀？我用insightface代码对训练集做了对齐，然后从你提供的云盘下载了lfw cfp adgdb30测试集，这样行吗？我担心的是训练集对齐方式和测试集对齐方式不一致

使用您的代码，我在lfw的得分可以达到99.6，但是实际应用场景会出现大批量的认错

do you did face alignment?

Visdom successfully connected to server '>' not supported between instances of 'float' and 'NoneType' Traceback (most recent call last): File "e:/AI-pro/Insightface/Face_Pytorch-master/train_my.py", line 263, in <module> train(args) File "e:/AI-pro/Insightface/Face_Pytorch-master/train_my.py", line 155, in train total_loss.backward() File "C:\Users\Administrator\Anaconda3\lib\site-packages\torch\tensor.py", line 107, in backward torch.autograd.backward(self, gradient, retain_graph, create_graph) File "C:\Users\Administrator\Anaconda3\lib\site-packages\torch\autograd\init.py", line 93, in backward allow_unreachable=True) # allow_unreachable flag RuntimeError: cuda runtime error (59) : device-side assert triggered at C:/w/1/s/windows/pytorch/aten/src\THC/generic/THCTensorMath.cu:26

I figured it out. Some images in my data is wrong, num_workers should be 0 in dataloader parameters.

[BUG/Help] <通过广告词训练之后感觉其他的回答认知出现了问题~>

全量或lora微调输入是问题和答案拼起来吗，输出仅仅是答案？

hub.Module加载预训练模型报错

我跑的这个文本分类，https://aistudio.baidu.com/aistudio/projectdetail/5794735?forkThirdPart=1 在AIstudio的环境里都跑不起来 trainer.train(train_dataset, epochs=10, batch_size=32, eval_dataset=dev_dataset, save_interval=1) # 配置训练参数，启动训练，并指定验证集 ---------------------------------------------------------------------------AttributeError Traceback (most recent call last)/tmp/ipykernel_2719/2900980602.py in ----> 1 trainer.train(train_dataset, epochs=10, batch_size=32, eval_dataset=dev_dataset, save_interval=1) # 配置训练参数，启动训练，并指定验证集 /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddlehub/finetune/trainer.py in train(self, train_dataset, epochs,...

RL related issues, code understanding , scores, rewards,ppo

I found self.reward_fn is: def reward_fn(samples: List[str], **kwargs): original_samples = [text.split("TL;DR:")[0] + "TL;DR: " for text in samples] original_samples = [text + post_summary_dict[text.strip()] for text in original_samples] original_scores = get_scores(original_samples)...

convert qwen-7b-chat , failed

the above solved, now new error: python llm_export.py --type Qwen-7B-Chat --path /mnt/LLM_Data/Qwen-7B-Chat --export_split --export_token --onnx_path /mnt/LLM_Data/Qwen-7B-Chat-onnx /home/ubuntu/anaconda3/envs/modelscope/lib/python3.10/site-packages/transformers/utils/generic.py:441: UserWarning: torch.utils._pytree._register_pytree_node is deprecated. Please use torch.utils._pytree.register_pytree_node instead. _torch_pytree._register_pytree_node( /home/ubuntu/anaconda3/envs/modelscope/lib/python3.10/site-packages/transformers/utils/generic.py:309: UserWarning: torch.utils._pytree._register_pytree_node is...

convert qwen-7b-chat , failed

上面问题没了，现在问题是报段错误

OOM despite ZeRO stage 3

what os version / cpu model/ main board do you use?