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

add new fairseq_pretraining function for starting from checkpoint #255

Merged
merged 4 commits into from
Jan 7, 2025
Merged

add new fairseq_pretraining function for starting from checkpoint #255

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
3464e55
add new fairseq_pretraining job with starting from cp
AndreasPlt Oct 22, 2024
2845bdd
use fairseq param for start checkpoint
AndreasPlt Dec 3, 2024
11cb4a8
simplify
AndreasPlt Dec 18, 2024
9b0660a
change str to tk.Path
AndreasPlt Jan 7, 2025
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
6 changes: 5 additions & 1 deletion ...experiments/librispeech/librispeech_960_pretraining/wav2vec2/config_02_fairseq_phoneme.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -63,7 +63,7 @@ def get_fairseq_root(commit="e4a2e4e93efbcbaaae52a17ae6600beb2083fb33", fairseq_
return fairseq_root


def run_fairseq_pretraining(exp_name, commit, python_exe_hash_overwrite=None, **kwargs):
def run_fairseq_pretraining(exp_name, commit, python_exe_hash_overwrite=None, checkpoint=None, **kwargs):
"""
Runs a FairseqHydraTrainingJob to pretrain a wav2vec 2.0 model.

Expand All @@ -73,6 +73,8 @@ def run_fairseq_pretraining(exp_name, commit, python_exe_hash_overwrite=None, **
python_exe_hash_overwrite (Optional[str]): The hash overwrite for the fairseq_python_exe to use.
It should only be used to achieve compatibility with the previous setup structure and should be ignored
in all other cases.
checkpoint (Optional[tk.Path]): The path to the checkpoint to start from. If None, the training will start
from scratch.
**kwargs: Additional arguments to pass to the job. These will be used to overwrite the model configuration.
"""
# job requirements
Expand All @@ -93,6 +95,8 @@ def run_fairseq_pretraining(exp_name, commit, python_exe_hash_overwrite=None, **
# generate config
fairseq_args = get_fairseq_args(num_gpus=num_gpus)
fairseq_args["task"]["alignment"] = alignment
if checkpoint is not None:
fairseq_args["checkpoint"]["continue_once"] = checkpoint
for k, v in kwargs.items():
fairseq_args["model"][k] = v
fairseq_config = FairseqHydraConfig(fairseq_args)
Expand Down
Loading