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

BCOTrainer conversational dataset support #2107

Merged
merged 8 commits into from
Sep 24, 2024
+27 −148
Merged

BCOTrainer conversational dataset support #2107

Changes from 1 commit
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
641d7a2
update test
qgallouedec Sep 24, 2024
351ed55
maybe_apply_chat_template
qgallouedec Sep 24, 2024
d273e43
simplify bco example
qgallouedec Sep 24, 2024
8cc574c
Update documentation
qgallouedec Sep 24, 2024
0058f47
Update examples/scripts/bco.py
qgallouedec Sep 24, 2024
d545d0d
Update docs/source/bco_trainer.mdx
qgallouedec Sep 24, 2024
53c72ad
Merge branch 'main' into bco-conversational-dataset
qgallouedec Sep 24, 2024
88cf41a
Merge branch 'main' into bco-conversational-dataset
qgallouedec Sep 24, 2024
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
Next Next commit
update test
  • Loading branch information
@qgallouedec
qgallouedec committed Sep 24, 2024
commit 641d7a24a69952a008671cae02d980fe99fe6d69
20 changes: 7 additions & 13 deletions tests/test_bco_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -49,19 +49,19 @@ def setUp(self):

@parameterized.expand(
[
["gpt2", True, True],
["gpt2", True, False],
["gpt2", False, True],
["gpt2", False, False],
["gpt2", True, True, "standard_unpaired_preference"],
["gpt2", True, False, "standard_unpaired_preference"],
["gpt2", False, True, "standard_unpaired_preference"],
["gpt2", False, False, "standard_unpaired_preference"],
["gpt2", True, True, "conversational_unpaired_preference"],
]
)
def test_bco_trainer(self, name, pre_compute, eval_dataset):
def test_bco_trainer(self, name, pre_compute, eval_dataset, config_name):
with tempfile.TemporaryDirectory() as tmp_dir:
training_args = BCOConfig(
output_dir=tmp_dir,
per_device_train_batch_size=2,
max_steps=3,
remove_unused_columns=False,
gradient_accumulation_steps=1,
learning_rate=9e-1,
eval_strategy="steps",
Expand All @@ -70,7 +70,7 @@ def test_bco_trainer(self, name, pre_compute, eval_dataset):
report_to="none",
)

dummy_dataset = load_dataset("trl-internal-testing/zen", "standard_unpaired_preference")
dummy_dataset = load_dataset("trl-internal-testing/zen", config_name)

if name == "gpt2":
model = self.model
Expand Down Expand Up @@ -129,7 +129,6 @@ def test_tokenize_and_process_tokens(self):
output_dir=tmp_dir,
per_device_train_batch_size=2,
max_steps=3,
remove_unused_columns=False,
gradient_accumulation_steps=1,
learning_rate=9e-1,
eval_strategy="steps",
Expand Down Expand Up @@ -192,7 +191,6 @@ def test_bco_trainer_without_providing_ref_model(self):
output_dir=tmp_dir,
per_device_train_batch_size=2,
max_steps=3,
remove_unused_columns=False,
gradient_accumulation_steps=4,
learning_rate=9e-1,
eval_strategy="steps",
Expand Down Expand Up @@ -230,7 +228,6 @@ def test_bco_trainer_udm(self):
output_dir=tmp_dir,
per_device_train_batch_size=2,
max_steps=3,
remove_unused_columns=False,
gradient_accumulation_steps=4,
learning_rate=9e-1,
eval_strategy="steps",
Expand Down Expand Up @@ -289,7 +286,6 @@ def test_bco_trainer_without_providing_ref_model_with_lora(self):
output_dir=tmp_dir,
per_device_train_batch_size=2,
max_steps=3,
remove_unused_columns=False,
gradient_accumulation_steps=4,
learning_rate=9e-1,
eval_strategy="steps",
Expand Down Expand Up @@ -330,7 +326,6 @@ def test_bco_trainer_generate_during_eval_no_wandb(self):
output_dir=tmp_dir,
per_device_train_batch_size=2,
max_steps=3,
remove_unused_columns=False,
gradient_accumulation_steps=1,
learning_rate=9e-1,
eval_strategy="steps",
Expand Down Expand Up @@ -376,7 +371,6 @@ def test_bco_lora_save(self):
output_dir=tmp_dir,
per_device_train_batch_size=2,
max_steps=3,
remove_unused_columns=False,
gradient_accumulation_steps=4,
learning_rate=9e-1,
eval_strategy="steps",
Expand Down