Basic CircleCI (ml-explore#449)

* basic style checks for circleci

* format

* fix config

.circleci/config.yml

@@ -0,0 +1,35 @@
+version: 2.1
+
+jobs:
+  linux_build_and_test:
+    docker:
+      - image: cimg/python:3.9
+
+    steps:
+      - checkout
+      - run:
+          name: Run style checks
+          command: |
+            pip install pre-commit
+            pre-commit run --all
+            if ! git diff --quiet; then echo 'Style checks failed, please install pre-commit and run pre-commit run --all and push the change'; exit 1; fi
+
+workflows:
+  build_and_test:
+    when:
+      matches:
+        pattern: "^(?!pull/)[-\\w]+$"
+        value: << pipeline.git.branch >>
+    jobs:
+      - linux_build_and_test
+
+  prb:
+    when:
+      matches:
+        pattern: "^pull/\\d+(/head)?$"
+        value: << pipeline.git.branch >>
+    jobs:
+      - hold:
+          type: approval
+      - linux_build_and_test:
+          requires: [ hold ]

llms/mlx_lm/tuner/trainer.py

@@ -99,7 +99,7 @@ def evaluate(
     num_batches,
     max_seq_length=2048,
     loss: callable = default_loss,
-    iterate_batches: callable = iterate_batches
+    iterate_batches: callable = iterate_batches,
 ):
     all_losses = []
     ntokens = 0
@@ -121,7 +121,14 @@ def evaluate(
 
 class TrainingCallback:
 
-    def on_train_loss_report(self, steps: int, loss: float, it_sec: float, tokens_sec: float, trained_tokens: int):
+    def on_train_loss_report(
+        self,
+        steps: int,
+        loss: float,
+        it_sec: float,
+        tokens_sec: float,
+        trained_tokens: int,
+    ):
         """Called to report training loss at specified intervals."""
         pass
 
@@ -193,7 +200,9 @@ def train(
             )
 
             if training_callback is not None:
-                training_callback.on_train_loss_report(it + 1, train_loss, it_sec, tokens_sec, trained_tokens)
+                training_callback.on_train_loss_report(
+                    it + 1, train_loss, it_sec, tokens_sec, trained_tokens
+                )
 
             losses = []
             n_tokens = 0
@@ -210,7 +219,7 @@ def train(
                 batch_size=args.batch_size,
                 num_batches=args.val_batches,
                 max_seq_length=args.max_seq_length,
-                iterate_batches=iterate_batches
+                iterate_batches=iterate_batches,
             )
             val_time = time.perf_counter() - stop
             print(