wanghao comments

Results 4 comments of


                                            wanghao

运行cuda报错，显示内存溢出

你这个问题比较复杂，建议重装显卡驱动。

hi，can you release 7B ckpt in first stage？

哈哈抱歉，我详细描述一下我的问题。我使用这里公布的MiniGPT-4 checkpoint(https://drive.google.com/file/d/1u9FRRBB3VovP1HxCAlpD9Lw4t4P6-Yq8/view?usp=share_link)直接做第二阶段的训练时出现这个维度错误：size mismatch for llama_proj.weight: copying a param with shape torch.Size([5120, 768]) from checkpoint, the shape in current model is torch.Size([4096, 768]). 我猜测这边给出的模型权重是13B llama训练的，所以想要7B训练的MiniGPT-4 checkpoint。谢谢。

The result of direct inference without using VLLM is wrong, is it a problem with the model?

我也遇到了这个问题，非常的奇怪。请问有解决的思路吗？

The result of direct inference without using VLLM is wrong, is it a problem with the model?

I had the same problem, and it was very strange. Is there a solution?