add

app.py

@@ -3,7 +3,7 @@ import torch
 from transformers import AutoTokenizer, AutoModelForCausalLM
 import gradio as gr
 from peft import PeftModel
-import spaces  # Ensure spaces is imported
+import spaces
 
 # Define the base model ID
 base_model_id = "meta-llama/Llama-2-13b-hf"
@@ -17,7 +17,7 @@ if not huggingface_token:
 base_model = AutoModelForCausalLM.from_pretrained(
     base_model_id,  
     trust_remote_code=True,
-    token=huggingface_token  # Use the token parameter
+    use_auth_token=huggingface_token  # Use the correct parameter
 ).to("cuda")  # Move model to CUDA
 
 # Load the tokenizer
@@ -25,7 +25,7 @@ tokenizer = AutoTokenizer.from_pretrained(
     base_model_id, 
     add_bos_token=True, 
     trust_remote_code=True, 
-    token=huggingface_token
+    use_auth_token=huggingface_token
 )
 
 # Load the fine-tuned model and move to CUDA
@@ -54,6 +54,9 @@ def generate_skills(job_description):
     else:
         skills_text = generated_text[skills_start_index:].strip()
 
+    # Clear CUDA memory
+    torch.cuda.empty_cache()
+
     return skills_text
 
 # Define the Gradio interface
@@ -61,4 +64,4 @@ inputs = gr.Textbox(lines=10, label="Job description:", placeholder="Enter or pa
 outputs = gr.Textbox(label="Required skills:", placeholder="The required skills will be displayed here...")
 
 gr.Interface(fn=generate_skills, inputs=inputs, outputs=outputs, title="Job Skills Analysis",
-             description="Paste the job description in the text box below and the model will show the required skills for candidates.").launch()
+             description="Paste the job description in the text box below and the model will show the required skills for candidates.").launch(share=True)