Vishal24/BCG_adapter_v1

Model Card for Model ID

Model Details

Model Description

• Developed by: [More Information Needed]
• Funded by [optional]: [More Information Needed]
• Shared by [optional]: [More Information Needed]
• Model type: [More Information Needed]
• Language(s) (NLP): [More Information Needed]
• License: [More Information Needed]
• Finetuned from model [optional]: [More Information Needed]

Infrence Function

for branded or generic

def generate1(keyword):

  prompt = f"""[INST] Annotate the keyword into branded or generic.[/INST]

      [KW] {keyword} [/KW]
      
      response ###"""
  print("Prompt:")
  print(prompt)
  encoding = tokenizer(prompt, return_tensors="pt").to("cuda:0")
  output = model.generate(input_ids=encoding.input_ids,
                          attention_mask=encoding.attention_mask,
                          max_new_tokens=200,
                          do_sample=True,
                          temperature=0.9,
                          eos_token_id=tokenizer.eos_token_id,
                          top_p=0.9,
                         repetition_penalty=1.2)

  print()
  # Subtract the length of input_ids from output to get only the model's response
  output_text = tokenizer.decode(output[0, len(encoding.input_ids[0]):], skip_special_tokens=False)
  output_text = re.sub('\n+', '\n', output_text)  # remove excessive newline characters
  print("Generated Assistant Response:")
  print(output_text)

  
  return output_text

for brand name

def generate2(lista,keyword):
  
  prompt = f"""[INST] Extract the brand of the keyword from the given list if present.[/INST]

      [KW] {keyword} [/KW]

      [LIST] {lista} [/LIST]

      
      response ###"""
  print("Prompt:")
  print(prompt)
  encoding = tokenizer(prompt, return_tensors="pt").to("cuda:0")
  output = model.generate(input_ids=encoding.input_ids,
                          attention_mask=encoding.attention_mask,
                          max_new_tokens=200,
                          do_sample=True,
                          temperature=0.9,
                          eos_token_id=tokenizer.eos_token_id,
                          top_p=0.9,
                         repetition_penalty=1.2)

  print()
  # Subtract the length of input_ids from output to get only the model's response
  output_text = tokenizer.decode(output[0, len(encoding.input_ids[0]):], skip_special_tokens=False)
  output_text = re.sub('\n+', '\n', output_text)  # remove excessive newline characters
  print("Generated Assistant Response:")
  return output_text