Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Miscellaneous fixes to automatic tests #35

Merged
merged 12 commits into from
Jul 22, 2022
Merged

Miscellaneous fixes to automatic tests #35

Changes from 1 commit
Commits
Show all changes
12 commits
Select commit Hold shift + click to select a range
c322865
Fix branch name in CI
justheuristic Jul 22, 2022
39dd3d9
wait for the first server to initialize DHT
justheuristic Jul 22, 2022
375f5ad
[WIP]
justheuristic Jul 22, 2022
6583763
remove debugprint
justheuristic Jul 22, 2022
aee295f
de-flappify: give servers more time to download the model
justheuristic Jul 22, 2022
dfe57eb
set dtype='auto'
justheuristic Jul 22, 2022
05fed96
remove torch_dtype on converted model
justheuristic Jul 22, 2022
3f048e5
undo half
justheuristic Jul 22, 2022
996104a
[WIP] disable conversion
justheuristic Jul 22, 2022
dc277b3
minimize peak CPU memory usage
justheuristic Jul 22, 2022
84532ee
revert debug changes
justheuristic Jul 22, 2022
abd3da6
Update .github/workflows/run-tests.yaml
dbaranchuk Jul 22, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
remove torch_dtype on converted model
  • Loading branch information
@justheuristic
justheuristic committed Jul 22, 2022
commit 05fed964a3ca963fdb39c8dc6ecdb0680421c958
2 changes: 1 addition & 1 deletion tests/test_full_model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,7 +13,7 @@
@pytest.mark.forked
def test_full_model_exact_match(atol_forward=1e-3, atol_inference=1e-3):
tokenizer = transformers.BloomTokenizerFast.from_pretrained(MODEL_NAME)
model = DistributedBloomForCausalLM.from_pretrained(MODEL_NAME, initial_peers=INITIAL_PEERS, torch_dtype="auto")
model = DistributedBloomForCausalLM.from_pretrained(MODEL_NAME, initial_peers=INITIAL_PEERS)
assert isinstance(model, DistributedBloomForCausalLM)
assert len(model.transformer.h) == model.config.n_layer

Expand Down