Update dataset loading and subprocess command in

train_llm.py

@@ -12,7 +12,7 @@ from tqdm.notebook import tqdm
 
 # Load dataset
 dataset_name = 'ai-aerospace/ams_data_train_generic_v0.1_100'
-dataset=load_dataset(dataset_name)
+dataset=load_dataset(dataset_name,cache_dir='.')
 
 # Write dataset files into data directory
 data_directory = '../fine_tune_data/'
@@ -46,10 +46,11 @@ os.environ["train_data"] = train_data
 os.environ["validation_data"] = validation_data
 
 # Set .venv and execute the autotrain script
+# To see all parameters: autotrain llm --help
 # !autotrain llm --train --project_name my-llm --model TinyLlama/TinyLlama-1.1B-Chat-v0.1 --data_path . --use-peft --use_int4 --learning_rate 2e-4 --train_batch_size 6 --num_train_epochs 3 --trainer sft
 # The training dataset to be used must be called training.csv and be located in the data_path folder.
 command="""
-source ../.venv/bin/activate && autotrain llm --train \
+autotrain llm --train \
     --project_name ${project_name} \
     --model ${model_name} \
     --data_path ../fine_tune_data \
@@ -66,4 +67,4 @@ source ../.venv/bin/activate && autotrain llm --train \
 """
 
 # Use subprocess.run() to execute the command
-subprocess.run(command, shell=True, check=True, env=os.environ)
+subprocess.run(command, shell=True, check=True)