AutomaTikZ

AutomaTikZ copied to clipboard

example/train.py RuntimeError: expected mat1 and mat2 to have the same dtype, but got: float != c10::Half.

4

It seems there are some precision error when I run **example/train.py**. Could you help me with this？ I try set llama "torch_dtype": "float**16**", into "torch_dtype": "float**32**".. It is not helpful...

JasonLLLLLLLLLLL

Can I continue training from a checkpoint?

1

It seems it can get the last checkpoint in **train/llama.py**. But the loss seems to start over again(at 1.6). It should be 0.2 at this checkpoint. ``` {'loss': 1.6927, 'learning_rate':...

JasonLLLLLLLLLLL

run examples/train.py with two 3090 GPUS?

1

How can I run the program with two GPUS? Do I need to add some code to support two GPUS? Hopefully you can give me some instructions. Thanks!

AdjugateMatrix

CUDA out of memory. 4090. train.py

3

when I run **examples/train.py** with a 4090(24GB). it shows cuda out memeory. May I ask what GPU do you use when run train.py? Thanks! torch.cuda.OutOfMemoryError: CUDA out of memory. Tried...

JasonLLLLLLLLLLL