in src/autotrain/parser.py [0:0]
def _parse_config(self):
params = {
"model": self.config["base_model"],
"project_name": self.config["project_name"],
}
params["data_path"] = self.config["data"]["path"]
if self.task == "lm_training":
params["chat_template"] = self.config["data"]["chat_template"]
if "-" in self.config["task"]:
params["trainer"] = self.config["task"].split("-")[1]
if params["trainer"] == "generic":
params["trainer"] = "default"
if params["trainer"] not in ["sft", "orpo", "dpo", "reward", "default"]:
raise ValueError("Invalid LLM training task")
if self.task == "sentence_transformers":
params["trainer"] = self.config["task"].split(":")[1]
if self.task == "vlm":
params["trainer"] = self.config["task"].split(":")[1]
for k, v in self.config["data"]["column_mapping"].items():
params[k] = v
params["train_split"] = self.config["data"]["train_split"]
params["valid_split"] = self.config["data"]["valid_split"]
params["log"] = self.config["log"]
if "hub" in self.config:
params["username"] = self.config["hub"]["username"]
params["token"] = self.config["hub"]["token"]
params["push_to_hub"] = self.config["hub"]["push_to_hub"]
else:
params["username"] = None
params["token"] = None
params["push_to_hub"] = False
if params["username"]:
if params["username"].startswith("${"):
params["username"] = os.environ.get(params["username"][2:-1])
if params["token"]:
if params["token"].startswith("${"):
params["token"] = os.environ.get(params["token"][2:-1])
other_params = self.config.get("params")
if other_params:
params.update(other_params)
return params