I tried to use dcp+fsdp for training, but I got an error when executing optimizer.step(). I found the relevant unit test in pytorch: test_e2e_save_and_load.py ...
from vllm.model_executor.layers.quantization.utils.fp8_utils import W8A8BlockFp8LinearOp from vllm.model_executor.layers.quantization.utils.quant_utils import ( ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results