diff --git a/mlx_lm/LORA.md b/mlx_lm/LORA.md
index 2825f48..514a679 100644
--- a/mlx_lm/LORA.md
+++ b/mlx_lm/LORA.md
@@ -66,9 +66,10 @@ mlx_lm.lora \
 To fine-tune the full model weights, add the `--fine-tune-type full` flag.
 Currently supported fine-tuning types are `lora` (default), `dora`, and `full`.
 
-The `--data` argument must specify a path to a `train.jsonl`, `valid.jsonl`
-when using `--train` and a path to a `test.jsonl` when using `--test`. For more
-details on the data format see the section on [Data](#Data).
+The `--data` argument must specify a path to a `train.jsonl` when using
+`--train` and a path to a `test.jsonl` when using `--test`. A `valid.jsonl` is
+optional; if provided, validation loss will be reported during training. For
+more details on the data format see the section on [Data](#Data).
 
 For example, to fine-tune a Mistral 7B you can use `--model
 mistralai/Mistral-7B-v0.1`.
@@ -184,9 +185,10 @@ Face.
 
 ### Local Datasets
 
-For fine-tuning (`--train`), the data loader expects a `train.jsonl` and a
-`valid.jsonl` to be in the data directory. For evaluation (`--test`), the data
-loader expects a `test.jsonl` in the data directory. 
+For fine-tuning (`--train`), the data loader expects a `train.jsonl` to be in
+the data directory. A `valid.jsonl` is optional; if present, validation loss
+will be reported periodically during training. For evaluation (`--test`), the
+data loader expects a `test.jsonl` in the data directory. 
 
 Currently, `*.jsonl` files support `chat`, `tools`, `completions`, and `text`
 data formats. Here are examples of these formats:
diff --git a/mlx_lm/tuner/datasets.py b/mlx_lm/tuner/datasets.py
index b8795d5..2c40d2d 100644
--- a/mlx_lm/tuner/datasets.py
+++ b/mlx_lm/tuner/datasets.py
@@ -322,8 +322,8 @@ def load_dataset(args, tokenizer: PreTrainedTokenizer):
             "Training set not found or empty. Must provide training set for fine-tuning."
         )
     if args.train and len(valid) == 0:
-        raise ValueError(
-            "Validation set not found or empty. Must provide validation set for fine-tuning."
+        print(
+            "Warning: Validation set not found or empty. Training will proceed without validation."
         )
     if args.test and len(test) == 0:
         raise ValueError(
diff --git a/mlx_lm/tuner/trainer.py b/mlx_lm/tuner/trainer.py
index 7b18c7f..7425498 100644
--- a/mlx_lm/tuner/trainer.py
+++ b/mlx_lm/tuner/trainer.py
@@ -205,7 +205,7 @@ def train(
     model,
     optimizer,
     train_dataset,
-    val_dataset,
+    val_dataset=None,
     args: TrainingArgs = TrainingArgs(),
     loss: callable = default_loss,
     iterate_batches: callable = iterate_batches,
@@ -269,7 +269,9 @@ def train(
         tic = time.perf_counter()
         # Report validation loss if needed, the first validation loss
         # is always measured before any training.
-        if it == 1 or it % args.steps_per_eval == 0 or it == args.iters:
+        if val_dataset and (
+            it == 1 or it % args.steps_per_eval == 0 or it == args.iters
+        ):
             tic = time.perf_counter()
             val_loss = evaluate(
                 model=model,