Gosula commited on
Commit
baa9f19
1 Parent(s): 5b48e40

Create app.py

Browse files
Files changed (1) hide show
  1. app.py +26 -0
app.py ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import gradio as gr
2
+
3
+ from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
4
+ import torch
5
+ from peft import LoraConfig, PeftModel
6
+
7
+ base_model_name = "microsoft/phi-2"
8
+ new_model = "./checkpoint_360"
9
+
10
+ model = AutoModelForCausalLM.from_pretrained( "microsoft/phi-2", trust_remote_code=True)
11
+ model.config.use_cache = False
12
+ model.load_adapter(new_model)
13
+
14
+ tokenizer = AutoTokenizer.from_pretrained(base_model_name, trust_remote_code=True)
15
+ tokenizer.pad_token = tokenizer.eos_token
16
+ tokenizer.padding_side = "right"
17
+
18
+ def QLoRA_Chatgpt(prompt):
19
+ print(prompt)
20
+ pipe = pipeline(task="text-generation", model=model, tokenizer=tokenizer, max_length=200)
21
+ result = pipe(f"<s>[INST] {prompt} [/INST]")
22
+ return(result[0]['generated_text'])
23
+ # return "Hello " + name + "!!"
24
+
25
+ iface = gr.Interface(fn=QLoRA_Chatgpt, inputs=gr.Textbox("how can help you today", label='prompt'), outputs=gr.Textbox(label='Generated-output'))
26
+ iface.launch(share=True)