[pre-commit.ci] pre-commit autoupdate (#532)

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

.pre-commit-config.yaml

@@ -21,13 +21,13 @@ repos:
             exclude: ^experiments/configs
 
     - repo: https://github.com/psf/black
-      rev: 23.12.1
+      rev: 24.3.0
       hooks:
           - id: black
             language_version: python3 # Should be a command that runs python3.6+
 
     - repo: https://github.com/PyCQA/flake8
-      rev: 6.1.0
+      rev: 7.0.0
       hooks:
           - id: flake8
             args: [--count, --show-source, --statistics]

data/check_pandas.py

@@ -8,6 +8,7 @@
 This script has a command line interface. You can run it using `python check_pandas <data_dir>`,
 where `<data_dir>` points to a nested set of directories with `data_clean.csv` files.
 """
+
 import os
 from glob import glob
 from pathlib import Path

data/check_smiles_split.py

@@ -6,6 +6,7 @@
 This script uses dask. This might cause some errors with mismatching data types,
 for which there are currently a few fallbacks.
 """
+
 import os
 from glob import glob
 from pathlib import Path

data/natural/preprocess_europepmc.py

@@ -5,6 +5,7 @@
 
 Before running this scripts, the filepaths need to be changed.
 """
+
 import json
 import os
 import re

data/natural/preprocess_msds.py

@@ -3,6 +3,7 @@
 
 You need to change filepaths before running this script
 """
+
 import json
 import os
 

data/natural/preprocess_nougat.py

@@ -6,6 +6,7 @@
 
 The filepaths need to be updated before running the script.
 """
+
 import glob
 import json
 import os

data/postprocess_split.py

@@ -5,6 +5,7 @@
 
 This script needs to be run after the splitting script.
 """
+
 import os
 from glob import glob
 from pathlib import Path

data/tabular/check_smiles_split.py

@@ -1,4 +1,5 @@
 """This script checks for data leakage in the splits of a tabular dataset."""
+
 import os
 from glob import glob
 from pathlib import Path

data/tabular/odd_one_out/transform.py

@@ -104,9 +104,11 @@ def transform_dataset(dataset, n_permutations):
         "smi_4": smis[smi_idx_arr[:, 3]],
         "odd_one_out_idx": odd_one_out_idx,
         "odd_one_out_mol": [
-            smis[smi_idx_arr[i, int(odd_one_out_idx[i])]]
-            if not np.isnan(odd_one_out_idx[i])
-            else np.nan
+            (
+                smis[smi_idx_arr[i, int(odd_one_out_idx[i])]]
+                if not np.isnan(odd_one_out_idx[i])
+                else np.nan
+            )
             for i in range(len(odd_one_out_idx))
         ],
         # "similarity_list": similarity_list,

data/tabular/train_test_split.py

@@ -20,6 +20,7 @@
     - Some CSV files contain complicated strings. We cannot parse them in a chunked manner.
         In this case, we set blocksize=None and read the whole file into memory.
 """
+
 import logging
 import os
 import random

data/text_sampling/text_sampling.py

@@ -942,9 +942,9 @@ def export(self, fn_suffix: str = None):
                     inplace=True,
                 )
                 if self.multiple_choice_benchmarking_templates:
-                    df_out[
-                        ["output", "answer_choices", "correct_output_index"]
-                    ] = df_out["output"].str.split(pat="<MC>", n=2, expand=True)
+                    df_out[["output", "answer_choices", "correct_output_index"]] = (
+                        df_out["output"].str.split(pat="<MC>", n=2, expand=True)
+                    )
                     df_out["answer_choices"] = df_out["answer_choices"].apply(
                         lambda x: x.split("|")
                     )
@@ -982,15 +982,15 @@ def export(self, fn_suffix: str = None):
                     os.makedirs(output_path_dir, exist_ok=True)
                     output_path = output_path_dir + f"{split}.jsonl"
 
-                    lm_eval_yaml_template_multiple_choice[
-                        "task"
-                    ] = self.path_data_dir.split("/")[-1]
-                    lm_eval_yaml_template_multiple_choice[
-                        "dataset_path"
-                    ] = output_path_dir
-                    lm_eval_yaml_template_multiple_choice[
-                        "dataset_name"
-                    ] = self.path_data_dir.split("/")[-1]
+                    lm_eval_yaml_template_multiple_choice["task"] = (
+                        self.path_data_dir.split("/")[-1]
+                    )
+                    lm_eval_yaml_template_multiple_choice["dataset_path"] = (
+                        output_path_dir
+                    )
+                    lm_eval_yaml_template_multiple_choice["dataset_name"] = (
+                        self.path_data_dir.split("/")[-1]
+                    )
 
                     fn_lm_eval_yaml = output_path_dir + "/config.yaml"
                     with open(fn_lm_eval_yaml, "w") as f:
@@ -1005,15 +1005,15 @@ def export(self, fn_suffix: str = None):
                     os.makedirs(output_path_dir, exist_ok=True)
                     output_path = output_path_dir + f"{split}_{fn_suffix}.jsonl"
 
-                    lm_eval_yaml_template_loglikelihood[
-                        "task"
-                    ] = self.path_data_dir.split("/")[-1]
-                    lm_eval_yaml_template_loglikelihood[
-                        "dataset_path"
-                    ] = output_path_dir
-                    lm_eval_yaml_template_loglikelihood[
-                        "dataset_name"
-                    ] = self.path_data_dir.split("/")[-1]
+                    lm_eval_yaml_template_loglikelihood["task"] = (
+                        self.path_data_dir.split("/")[-1]
+                    )
+                    lm_eval_yaml_template_loglikelihood["dataset_path"] = (
+                        output_path_dir
+                    )
+                    lm_eval_yaml_template_loglikelihood["dataset_name"] = (
+                        self.path_data_dir.split("/")[-1]
+                    )
 
                     fn_lm_eval_yaml = output_path_dir + "/config.yaml"
                     with open(fn_lm_eval_yaml, "w") as f:

data/train_test_split.py

@@ -20,6 +20,7 @@
     - Some CSV files contain complicated strings. We cannot parse them in a chunked manner.
         In this case, we set blocksize=None and read the whole file into memory.
 """
+
 import logging
 import os
 import random

experiments/data/merge_epmc_to_jsonl.py

@@ -7,6 +7,7 @@
 <dir>/2022_05_25/file2.jsonl
 ...
 """
+
 import multiprocessing
 import os
 from typing import List

experiments/data/prepare_gptneox_chemrxiv.py

@@ -5,6 +5,7 @@
 Example usage:
     python experiments/chem_data_prep.py /fsx/proj-chemnlp/data/ chemnlp/gpt-neox/
 """
+
 import argparse
 import os
 

experiments/data/prepare_hf_dataset.py

@@ -5,6 +5,7 @@
 Example Usage:
     python prepare_hf_dataset.py full_path/config.yml
 """
+
 import argparse
 import json
 import os

experiments/scripts/eval_create_batch_configs.py

@@ -20,9 +20,9 @@ def run(
     ]
 
     for model_name in model_names:
-        raw_config[
-            "model_args"
-        ] = f"pretrained={root_models_path}/{model_name}/{CHECKPOINT_DIR}"
+        raw_config["model_args"] = (
+            f"pretrained={root_models_path}/{model_name}/{CHECKPOINT_DIR}"
+        )
         raw_config["wandb_run_name"] = model_name
 
         with open(

experiments/scripts/run_tune.py

@@ -3,6 +3,7 @@
 
     Usage: python run_tune.py <path-to-config-yml>
 """
+
 import argparse
 import json
 import os
@@ -98,9 +99,9 @@ def run(config_path: str, config_overrides: Optional[Dict] = None) -> None:
     model_ref = getattr(transformers, config.model.base)
     model = model_ref.from_pretrained(
         pretrained_model_name_or_path=config.model.checkpoint_path or config.model.name,
-        revision=config.model.revision
-        if config.model.checkpoint_path is None
-        else None,
+        revision=(
+            config.model.revision if config.model.checkpoint_path is None else None
+        ),
     )
 
     if config.prompt_tuning.enabled:
@@ -171,9 +172,11 @@ def run(config_path: str, config_overrides: Optional[Dict] = None) -> None:
         **config.trainer.dict(exclude={"deepspeed_config", "restart_checkpoint"}),
         report_to="wandb" if config.wandb.enabled else "none",
         local_rank=local_rank,
-        deepspeed=CONFIG_DIR / f"deepspeed/{config.trainer.deepspeed_config}"
-        if config.trainer.deepspeed_config
-        else None,
+        deepspeed=(
+            CONFIG_DIR / f"deepspeed/{config.trainer.deepspeed_config}"
+            if config.trainer.deepspeed_config
+            else None
+        ),
     )
     print_zero_rank(local_rank, training_args)
 

src/chemnlp/data_val/config.py

@@ -10,9 +10,9 @@ class Data(BaseModel):
     path: Union[List[str], str]  # can be local or S3 directory
     validation_size: Union[List[float], float] = 0.05
     interleave_probs: Optional[List[float]] = None
-    sampling_criterion: Optional[
-        Literal["first_exhausted", "all_exhausted"]
-    ] = None  # as of v2.10.1
+    sampling_criterion: Optional[Literal["first_exhausted", "all_exhausted"]] = (
+        None  # as of v2.10.1
+    )
 
     @validator("validation_size")
     def small_positive_validation_sizes(cls, value_orig):

src/chemnlp/trainer.py

@@ -1,4 +1,5 @@
 """A custom trainer for modifying data sampling behaviour"""
+
 from typing import Optional
 
 import datasets