Upload app.py

app.py

@@ -3,7 +3,7 @@ import gradio as gr
 from transformers import AutoModelForCausalLM, AutoTokenizer
 import torch
 
-model_name = "Rallio67/chip_1.4B_instruct_alpha"
+model_name = "zirui3/gpt_1.4B_oa_instruct"
 tokenizer = AutoTokenizer.from_pretrained(model_name)
 
 chip_map= {
@@ -12,8 +12,8 @@ chip_map= {
     'gpt_neox.final_layer_norm': 0,
     'embed_out': 0
  }
-# model = AutoModelForCausalLM.from_pretrained(name, device_map=chip_map, torch_dtype=torch.float16, load_in_8bit=True)
-model = AutoModelForCausalLM.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(name, device_map=chip_map, torch_dtype=torch.float16, load_in_8bit=True)
+#model = AutoModelForCausalLM.from_pretrained(model_name)
 
 
 def predict(input, history=[], MAX_NEW_TOKENS = 500):
@@ -49,4 +49,4 @@ with gr.Blocks() as demo:
 
 if __name__ == "__main__":
     # demo.launch(debug=True, server_name="0.0.0.0", server_port=9991)
-    demo.launch()
+    demo.launch()