embed-train 2.0.0__tar.gz → 3.0.0__tar.gz

{embed_train-2.0.0 → embed_train-3.0.0}/CHANGELOG.md

@@ -1,3 +1,15 @@
+# [3.0.0](https://gitlab.com/efysent/agentic-core/embed-train/compare/v2.0.0...v3.0.0) (2026-05-09)
+
+
+* feat!: rename trainer configuration fields for consistency ([c3538f1](https://gitlab.com/efysent/agentic-core/embed-train/commit/c3538f1eadeeb708fa50962e648da000e471b1e9))
+
+
+### BREAKING CHANGES
+
+* Renamed training configuration fields across PyTorch and Hugging Face trainers:
+- config.lr -> config.learning_rate
+- config.batch_size -> config.per_device_train_batch_size (HF trainer only)
+
 # [2.0.0](https://gitlab.com/efysent/agentic-core/embed-train/compare/v1.0.0...v2.0.0) (2026-05-08)
 
 

{embed_train-2.0.0 → embed_train-3.0.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: embed-train
-Version: 2.0.0
+Version: 3.0.0
 Author-email: jalal <jalalkhaldi3@gmail.com>
 Requires-Python: <3.13,>=3.11
 Requires-Dist: accelerate<2.0.0,>=1.13.0

{embed_train-2.0.0 → embed_train-3.0.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "embed-train"
-version = "2.0.0"
+version = "3.0.0"
 description = ""
 authors = [
     { name = "jalal", email = "jalalkhaldi3@gmail.com" }

{embed_train-2.0.0 → embed_train-3.0.0}/src/embed_train/settings.py

@@ -82,7 +82,7 @@ class PyTorchTrainerSettings[
     num_epochs: int
     batch_size: int
     shuffle: bool
-    lr: float
+    learning_rate: float
     device: str
     save_every: int
     drop_last: bool
@@ -172,9 +172,9 @@ class SentenceTransformersTrainerSettings[
     tokenizer: TokenizerSettings
     loss: SentenceTransformerLoss
     pooling: Literal["cls", "mean_tokens", "max_tokens"]
-    batch_size: int
+    per_device_train_batch_size: int
     num_epochs: int
-    lr: float
+    learning_rate: float
     warmup_ratio: float
     eval_steps: int
     save_steps: int

{embed_train-2.0.0 → embed_train-3.0.0}/src/embed_train/train/trainers/hf/__init__.py

@@ -40,9 +40,10 @@ class SentenceTransformersTrainer[TCHFTrainRunner: "SentenceTransformersTrainerS
         warmup_steps = self._get_warmup_steps(train_dataset)
         args = SentenceTransformerTrainingArguments(
             output_dir=str(self.config.data_dir / f"checkpoints/{self._run_name()}"),
-            per_device_train_batch_size=self.config.batch_size,
+            per_device_train_batch_size=self.config.per_device_train_batch_size,
+            per_device_eval_batch_size=self.config.per_device_train_batch_size,
             num_train_epochs=self.config.num_epochs,
-            learning_rate=self.config.lr,
+            learning_rate=self.config.learning_rate,
             warmup_steps=warmup_steps,
             eval_strategy="steps",
             save_strategy="steps",
@@ -65,7 +66,7 @@ class SentenceTransformersTrainer[TCHFTrainRunner: "SentenceTransformersTrainerS
 
     def _get_warmup_steps(self, dataset: Dataset) -> float:
         train_size = len(dataset)
-        steps_per_epoch = train_size // self.config.batch_size
+        steps_per_epoch = train_size // self.config.per_device_train_batch_size
         total_steps = steps_per_epoch * self.config.num_epochs
         warmup_steps = int(total_steps * self.config.warmup_ratio)
         return warmup_steps

{embed_train-2.0.0 → embed_train-3.0.0}/src/embed_train/train/trainers/torch/__init__.py

@@ -146,7 +146,7 @@ class PyTorchTrainer[TCPyTorchTrainer: "PyTorchTrainerSettings[Any, Any, Any, An
     def _run_name(self) -> str:
         ts = datetime.now().strftime("%Y%m%d-%H%M%S")
         class_name = self.config.module_path.split(".")[-1].lower()
-        return f"{class_name}_bs{self.config.batch_size}_lr{self.config.lr}_{ts}"
+        return f"{class_name}_bs{self.config.batch_size}_lr{self.config.learning_rate}_{ts}"
 
     def _load_model(self) -> Model[Any]:
         return cast(
@@ -157,7 +157,7 @@ class PyTorchTrainer[TCPyTorchTrainer: "PyTorchTrainerSettings[Any, Any, Any, An
     def _load_optimizer(self) -> Optimizer:
         return torch.optim.AdamW(
             self.model.parameters(),
-            lr=self.config.lr,
+            lr=self.config.learning_rate,
         )
 
     def _load_loss(self) -> Loss[Any]:

{embed_train-2.0.0 → embed_train-3.0.0}/tests/fixtures/components.py

@@ -330,7 +330,7 @@ def build_pytorch_trainer_settings(tmp_path: Path, **overrides: Any) -> DummyPyT
         "num_epochs": 1,
         "batch_size": 2,
         "shuffle": True,
-        "lr": 0.01,
+        "learning_rate": 0.01,
         "device": "cpu",
         "save_every": 1,
         "drop_last": False,
@@ -432,9 +432,9 @@ def build_sentence_transformers_trainer_settings(**overrides: Any) -> SentenceTr
         tokenizer=build_tokenizer_settings(),
         loss=build_sentence_transformer_loss(),
         pooling="mean_tokens",
-        batch_size=2,
+        per_device_train_batch_size=2,
         num_epochs=1,
-        lr=0.01,
+        learning_rate=0.01,
         warmup_ratio=0.1,
         eval_steps=2,
         save_steps=4,

{embed_train-2.0.0 → embed_train-3.0.0}/uv.lock

@@ -459,7 +459,7 @@ wheels = [
 
 [[package]]
 name = "embed-train"
-version = "2.0.0"
+version = "3.0.0"
 source = { editable = "." }
 dependencies = [
     { name = "accelerate" },