add configs_dir

cohere-ai · Oct 30, 2024 · ace341e · ace341e
1 parent f9c1167
commit ace341e
Show file tree

Hide file tree

Showing 2 changed files with 9 additions and 9 deletions.
diff --git a/src/cohere_finetune/configs.py b/src/cohere_finetune/configs.py
@@ -161,12 +161,13 @@ def __init__(self, finetune_name: str) -> None:
         self.input_train_dir = os.path.join(self.root_dir, "input/data/training")
         self.input_eval_dir = os.path.join(self.root_dir, "input/data/evaluation")
 
-        # Finetune: paths of the preprocessed train & eval data and directories of checkpoints, logs, metrics (optional)
+        # Finetune: paths of the preprocessed train & eval data and directories of checkpoints, logs, configs, metrics (optional)
         self.finetune_dir = os.path.join(self.root_dir, "finetune")
         self.finetune_train_path = os.path.join(self.finetune_dir, "data/train.csv")
         self.finetune_eval_path = os.path.join(self.finetune_dir, "data/eval.csv")
         self.checkpoints_dir = os.path.join(self.finetune_dir, "checkpoints")
         self.logs_dir = os.path.join(self.finetune_dir, "logs")
+        self.configs_dir = os.path.join(self.finetune_dir, "configs")
         self.metrics_dir = os.path.join(self.finetune_dir, "metrics")
 
         # Output: directories of trained merged weights, trained adapter weights (optional), inference engine (optional)
@@ -184,10 +185,9 @@ def __init__(self, finetune_name: str) -> None:
         super().__init__(finetune_name)
 
         self.finetune_backends_parent_dir = os.path.dirname(__file__)
-        self.peft_template_parallel_configs_dir = \
-            f"{self.finetune_backends_parent_dir}/finetune_backends/cohere_peft/template_parallel_configs"
-        self.peft_parallel_config_dir = \
-            f"{self.finetune_backends_parent_dir}/finetune_backends/cohere_peft/parallel_config"
+        self.peft_template_parallel_configs_dir = os.path.join(
+            self.finetune_backends_parent_dir, "finetune_backends/cohere_peft/template_parallel_configs"
+        )
 
 
 def get_path_config(finetune_name: str) -> PathConfig:

diff --git a/src/cohere_finetune/train.py b/src/cohere_finetune/train.py
@@ -27,25 +27,25 @@ def train_with_peft(path_config: CoherePeftPathConfig, hyperparameters: Hyperpar
             use_postprocessed_model = False
             config_file_path = create_parallel_config_from_template(
                 os.path.join(path_config.peft_template_parallel_configs_dir, "fsdp_config.yaml"),
-                path_config.peft_parallel_config_dir,
+                path_config.configs_dir,
             )
         elif hyperparameters.parallel_strategy == "fsdp" and hyperparameters.use_4bit_quantization:
             use_postprocessed_model = False
             config_file_path = create_parallel_config_from_template(
                 os.path.join(path_config.peft_template_parallel_configs_dir, "fsdp_qlora_config.yaml"),
-                path_config.peft_parallel_config_dir,
+                path_config.configs_dir,
             )
         elif hyperparameters.parallel_strategy == "deepspeed" and not hyperparameters.use_4bit_quantization:
             use_postprocessed_model = True
             config_file_path = create_parallel_config_from_template(
                 os.path.join(path_config.peft_template_parallel_configs_dir, "deepspeed_z3_config.yaml"),
-                path_config.peft_parallel_config_dir,
+                path_config.configs_dir,
             )
         else:
             use_postprocessed_model = True
             config_file_path = create_parallel_config_from_template(
                 os.path.join(path_config.peft_template_parallel_configs_dir, "deepspeed_z3_qlora_config.yaml"),
-                path_config.peft_parallel_config_dir,
+                path_config.configs_dir,
             )
         peft_cmd = ["accelerate", "launch", "--config_file", config_file_path]
     else: