chore: lint

src/axolotl/cli/__init__.py

@@ -330,7 +330,7 @@ def load_datasets(
 def load_rl_datasets(
     *,
     cfg: DictDefault,
-    cli_args: TrainerCliArgs,
+    cli_args: TrainerCliArgs,  # pylint: disable=unused-argument
 ) -> TrainDatasetMeta:
     train_datasets: List[Any] = []
     for i, ds_cfg in enumerate(cfg.datasets):
@@ -340,7 +340,7 @@ def load_rl_datasets(
     # )
     eval_dataset = None
 
-    def argilla_apply_chatml(sample):
+    def argilla_apply_chatml(sample):  # pylint: disable=possibly-unused-variable
         if "system" in sample and sample["system"]:
             sample["prompt"] = (
                 f"<|im_start|>system\n{sample['system']}<|im_end|>\n"
@@ -354,7 +354,7 @@ def argilla_apply_chatml(sample):
         sample["rejected"] = f"{sample['rejected_response']}<|im_end|>"
         return sample
 
-    def intel_apply_chatml(sample):
+    def intel_apply_chatml(sample):  # pylint: disable=possibly-unused-variable
         if "system" in sample and sample["system"]:
             sample["prompt"] = (
                 f"<|im_start|>system\n{sample['system']}<|im_end|>\n"
@@ -368,7 +368,7 @@ def intel_apply_chatml(sample):
         sample["rejected"] = f"{sample['rejected']}<|im_end|>"
         return sample
 
-    def apply_chatml(sample):
+    def apply_chatml(sample):  # pylint: disable=possibly-unused-variable
         if "system" in sample and sample["system"]:
             sample["prompt"] = (
                 f"<|im_start|>system\n{sample['system']}<|im_end|>\n"
@@ -382,7 +382,7 @@ def apply_chatml(sample):
         sample["rejected"] = f"{sample['rejected']}<|im_end|>"
         return sample
 
-    def ultra_apply_chatml(sample):
+    def ultra_apply_chatml(sample):  # pylint: disable=possibly-unused-variable
         if "system" in sample and sample["system"]:
             sample["prompt"] = (
                 f"<|im_start|>system\n{sample['system']}<|im_end|>\n"
@@ -396,10 +396,10 @@ def ultra_apply_chatml(sample):
         sample["rejected"] = f"{sample['rejected'][1]['content']}<|im_end|>"
         return sample
 
-    for i, ds in enumerate(train_datasets):
+    for i, data_set in enumerate(train_datasets):
         _type = cfg.datasets[i]["type"]
-        fn = locals()[_type]
-        train_datasets[i] = ds.map(fn)
+        ds_type_fn = locals()[_type]
+        train_datasets[i] = data_set.map(ds_type_fn)
     train_dataset = concatenate_datasets(train_datasets)
 
     # eval_dataset = eval_dataset.map(intel_apply_chatml)

src/axolotl/core/trainers/trl.py

@@ -1,10 +1,21 @@
+"""
+module for TRL PPO training
+"""
 import torch
 from tqdm import tqdm
 from trl import PPOTrainer
 
 
 class TRLPPOTrainer(PPOTrainer):
-    def train(self, reward_pipe, resume_from_checkpoint=None):
+    """
+    wrapper for ppo trainer to handle customizations
+    """
+
+    def train(
+        self,
+        reward_pipe,
+        resume_from_checkpoint=None,  # pylint: disable=unused-argument
+    ):
         generation_kwargs = {
             "min_length": -1,
             "top_k": 0.0,
@@ -19,7 +30,9 @@ def train(self, reward_pipe, resume_from_checkpoint=None):
             "batch_size": 16,
         }
 
-        for epoch, batch in tqdm(enumerate(self.dataloader)):
+        for epoch, batch in tqdm(  # pylint: disable=unused-variable
+            enumerate(self.dataloader)
+        ):
             query_tensors = batch["input_ids"]
 
             # generate model response

tests/core/test_trainer_builder.py

@@ -1,8 +1,6 @@
 """
 unit tests for axolotl.core.trainer_builder
 """
-import unittest
-
 import pytest
 
 from axolotl.core.trainer_builder import HFDPOTrainerBuilder
@@ -45,7 +43,7 @@ def fixture_model(cfg, tokenizer):
     return load_model(cfg, tokenizer)
 
 
-class TestHFDPOTrainerBuilder(unittest.TestCase):
+class TestHFDPOTrainerBuilder:
     """
     TestCase class for DPO trainer builder
     """