Training model #11
Loading…
Reference in New Issue
Block a user
No description provided.
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Hello,
I try to quickly test the command python -m vall_e.models.ar_nar --yaml="./data/config.yaml" but it fails.
[rank0]: File "./vall_e/models/arch/llama.py", line 125, in forward
[rank0]: attn_output = torch.nn.functional.scaled_dot_product_attention(
[rank0]: RuntimeError: No available kernel. Aborting execution.
I am running on cuda 11.8 and torch-2.3.0+cu118. I also attempt to check on different torch version but the problem persists
Which version did you run successfully? Please suggest a solution to fix this error?
Thank you.
Oops, don't know how I never caught this. I must've made too much of an assumption that
sdpa
is available all the time (since it works on my V100s and I consider them old).Should be resolved in commit
949339a3fa
. If it doesn't, in./data/config.yaml
, set it toattention: eager
.