Update app.py

app.py

@@ -7,8 +7,8 @@ from gradio.themes.utils import colors, fonts, sizes
 from llama_cpp import Llama
 from huggingface_hub import hf_hub_download
 
-hf_hub_download(repo_id="Pi3141/alpaca-lora-13B-ggml", filename="ggml-model-q4_1.bin", local_dir=".")
-llm = Llama(model_path="./ggml-model-q4_1.bin", n_ctx=1048)
+hf_hub_download(repo_id="Pi3141/alpaca-lora-7B-ggml", filename="ggml-model-q4_1.bin", local_dir=".")
+llm = Llama(model_path="./ggml-model-q4_1.bin", n_threads=2)
 
 
 ins = '''Below is an instruction that describes a task. Write a response that appropriately completes the request.
@@ -28,9 +28,18 @@ theme = gr.themes.Monochrome(
 
 
 
+
+# def generate(instruction): 
+#     response = llm(ins.format(instruction))
+#     response = response['choices'][0]['text']
+#     result = ""
+#     for word in response.split(" "):
+#         result += word + " "
+#         yield result
+
 def generate(instruction): 
     result = ""
-    for x in llm(ins.format(instruction), stop=['### Instruction:', '### End'], stream=True, max_tokens=1024):
+    for x in llm(ins.format(instruction), stop=['### Instruction:', '### End'], stream=True):
         result += x['choices'][0]['text']
         yield result
 
@@ -105,7 +114,7 @@ with gr.Blocks(theme=seafoam, analytics_enabled=False, css=css) as demo:
         gr.Markdown(
             """ ## Alpaca-LoRa
             
-            is a 13B-parameter LLaMA model finetuned to follow instructions. It is trained on the Stanford Alpaca dataset and makes use of the Huggingface LLaMA implementation.
+            is a 7B-parameter LLaMA model finetuned to follow instructions. It is trained on the Stanford Alpaca dataset and makes use of the Huggingface LLaMA implementation.
             
             Type in the box below and click the button to generate answers to your most pressing questions!