Update README.md

README.md

@@ -1,6 +1,12 @@
 ---
 library_name: peft
 base_model: google-t5/t5-base
+license: apache-2.0
+language:
+- en
+- ja
+- ar
+pipeline_tag: text2text-generation
 ---
 
 # Model Card for Model ID
@@ -39,7 +45,44 @@ base_model: google-t5/t5-base
 
 ### Direct Use
 
-<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+`from peft import PeftModel
+
+model_id = 'google-t5/t5-base'
+
+bnb_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    load_4bit_use_double_quant=True,
+    bnb_4bit_quant_type="nf4",
+    bnb_4bit_compute_dtype=torch.bfloat16,
+)
+
+original_model = AutoModelForSeq2SeqLM.from_pretrained(model_id,quantization_config=bnb_config,device_map='auto')
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+
+tokenizer.pad_token = tokenizer.eos_token
+
+peft_model = PeftModel.from_pretrained(original_model, "bhuvanmdev/t5-base-news-describer")
+
+generation_config = peft_model.generation_config
+generation_config.do_sample = True
+generation_config.max_new_tokens = 100   # maxium no of token in output will get
+generation_config.temperature = 0.1
+generation_config.top_p = 0.8
+generation_config.num_return_sequences = 1
+generation_config.pad_token_id = tokenizer.eos_token_id
+generation_config.eos_token_id = tokenizer.eos_token_id
+generation_config.use_cache = True
+
+prompt = f"""Title: A big accidient occurs in luxemberg.""".strip()
+
+encoding = tokenizer(prompt, return_tensors="pt").to(device)
+with torch.inference_mode():
+    outputs = peft_model.generate(
+        input_ids=encoding.input_ids,
+        attention_mask=encoding.attention_mask,
+        generation_config=generation_config,
+    )
+print(tokenizer.decode(outputs[0], skip_special_tokens=True))`
 
 [More Information Needed]