Adding usage example. (#1)

- Adding usage example. (4bf47de4bec6aa6fc26000d9982c7c9a6a5be1e2)


Co-authored-by: palash sharma <palashsharma15@users.noreply.huggingface.co>

README.md

@@ -39,9 +39,27 @@ base_model: codellama/CodeLlama-7b-hf
 
 ### Direct Use
 
-<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
-
-[More Information Needed]
+```
+from peft import PeftModel, PeftConfig
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+config = PeftConfig.from_pretrained("hynky/codellama-7b-sft-lora-func-names-java-4bit")
+model = AutoModelForCausalLM.from_pretrained("codellama/CodeLlama-7b-hf",
+                                             torch_dtype='auto',
+                                             device_map='auto',
+                                             offload_folder="offload",
+                                             offload_state_dict = True)
+model = PeftModel.from_pretrained(model, "hynky/codellama-7b-sft-lora-func-names-java-4bit")
+
+def generate_code(sample, max_new_tokens=200):
+    batch = tokenizer(sample, return_tensors='pt').to(device)
+    
+    with torch.cuda.amp.autocast():
+        output_tokens = model.generate(**batch, max_new_tokens=max_new_tokens)
+    return tokenizer.decode(output_tokens[0], skip_special_tokens=True)
+
+print(generate_code("public class AddTwoIntegers("))
+```
 
 ### Downstream Use [optional]