ran precommit pipeline

model/model_training/README.md

@@ -4,8 +4,7 @@
 
 `pip install .`
 
-Run tests:
-`pytest .`
+Run tests: `pytest .`
 
 Start training SFT model
 

model/model_training/tests/test_datasets.py

@@ -3,7 +3,7 @@
 import pytest
 from custom_datasets import QA_DATASETS, SUMMARIZATION_DATASETS, get_one_dataset
 from custom_datasets.dialogue_collator import DialogueDataCollator
-from custom_datasets.prompt_dialogue import OAPrivate, PrivateInstructionTuning
+from custom_datasets.prompt_dialogue import OAPrivate
 from torch.utils.data import ConcatDataset, DataLoader
 from utils import get_tokenizer
 
@@ -20,7 +20,6 @@ def test_rl_sft_mode_switch():
 
 @pytest.mark.skip(reason="very slow")
 def test_all_datasets():
-
     qa_base = QA_DATASETS
     summarize_base = SUMMARIZATION_DATASETS
     others = ["webgpt", "soda", "joke", "explain_prosocial", "prosocial_dialogue"]
@@ -39,7 +38,6 @@ def test_all_datasets():
 
 @pytest.mark.skip(reason="very slow")
 def test_collate_fn():
-
     config = Namespace(cache_dir=".cache", model_name="Salesforce/codegen-2B-multi")
     tokenizer = get_tokenizer(config)
     collate_fn = DialogueDataCollator(tokenizer, max_length=620)

model/model_training/tests/test_oasst_dataset.py

@@ -1,7 +1,8 @@
 from argparse import Namespace
 
-from custom_datasets import get_one_dataset
 import pytest
+from custom_datasets import get_one_dataset
+
 
 @pytest.mark.skip(reason="cache not populated")
 def test_load_oasst_export_dataset():

model/reward/instructor/README.md

@@ -6,8 +6,7 @@ Trainer code based on huggingface. Compatible with deepspeed or accelerate
 
 `pip install .`
 
-Run tests:
-`pytest .`
+Run tests: `pytest .`
 
 Write or inherit a `configs/<config-name>.yml` file to store training
 configuration details.