refine

llama_stack/apis/post_training/post_training.py

@@ -30,7 +30,7 @@ class OptimizerType(Enum):
 @json_schema_type
 class DatasetFormat(Enum):
     instruct = "instruct"
-    chat = "chat"
+    chat = "dialog"
 
 
 @json_schema_type

llama_stack/providers/inline/post_training/common/validator.py

@@ -23,7 +23,7 @@ EXPECTED_DATASET_SCHEMA = {
             ColumnName.expected_answer.value: StringType(),
         }
     ],
-    "chat": [
+    "dialog": [
         {
             ColumnName.dialog.value: StringType(),
         }

llama_stack/providers/inline/post_training/torchtune/common/utils.py

@@ -49,7 +49,7 @@ MODEL_CONFIGS: Dict[str, ModelConfig] = {
 
 DATA_FORMATS: Dict[str, Transform] = {
     "instruct": InputOutputToMessages,
-    "chat": ShareGPTToMessages,
+    "dialog": ShareGPTToMessages,
 }
 
 

llama_stack/providers/inline/post_training/torchtune/datasets/format_adapter.py

@@ -30,6 +30,7 @@ def llama_stack_instruct_to_torchtune_instruct(
     input_message = input_messages[0]
 
     assert "content" in input_message, "content not found in input message"
+    input = input_message["content"]
     output = sample[ColumnName.expected_answer.value]
 
     return {

llama_stack/providers/inline/post_training/torchtune/recipes/lora_finetuning_single_device.py

@@ -135,7 +135,7 @@ class LoraFinetuningSingleDevice:
         self._data_format = training_config.data_config.data_format
         self._shuffle = training_config.data_config.shuffle
         self._batch_size = training_config.data_config.batch_size
-        self._training_on_input = training_config.data_config.training_on_input
+        self._train_on_input = training_config.data_config.train_on_input
 
         # this is important for debugging purpose
         self.max_steps_per_epoch = training_config.max_steps_per_epoch
@@ -367,7 +367,7 @@ class LoraFinetuningSingleDevice:
         data_transform = await utils.get_data_transform(self._data_format)
         ds = SFTDataset(
             rows,
-            message_transform=data_transform(train_on_input=self._training_on_input),
+            message_transform=data_transform(train_on_input=self._train_on_input),
             model_transform=tokenizer,
             dataset_type=self._data_format.value,
         )

llama_stack/templates/experimental-post-training/run.yaml

@@ -29,8 +29,8 @@ providers:
     provider_type: inline::basic
     config: {}
   datasetio:
-  - provider_id: huggingface-0
+  - provider_id: localfs
-    provider_type: remote::huggingface
+    provider_type: inline::localfs
     config: {}
   telemetry:
   - provider_id: meta-reference
@@ -68,23 +68,6 @@ metadata_store:
 models: []
 shields: []
 memory_banks: []
-datasets:
+datasets: []
-  - dataset_id: alpaca
-    provider_id: huggingface-0
-    url:
-      uri: https://huggingface.co/datasets/tatsu-lab/alpaca
-    metadata:
-      path: tatsu-lab/alpaca
-      name:
-      split: train
-    dataset_schema:
-      instruction:
-        type: string
-      input:
-        type: string
-      output:
-        type: string
-      text:
-        type: string
 scoring_fns: []
 eval_tasks: []