Spaces:

liamebs
/

LLMfromCPUoptimized-GGML-format

Sleeping

App Files Files Community

LLMfromCPUoptimized-GGML-format / app.py

liamebs

Update app.py

3fd6e5c 9 months ago

raw history blame contribute delete

No virus

1.88 kB

	#!/usr/bin/env python
	# coding: utf-8

	# In[ ]:


	import os
	import urllib.request
	import gradio as gr
	from llama_cpp import Llama


	def download_file(file_link, filename):
	# Checks if the file already exists before downloading
	if not os.path.isfile(filename):
	urllib.request.urlretrieve(file_link, filename)
	print("File downloaded successfully.")
	else:
	print("File already exists.")


	# Dowloading GGML model from HuggingFace
	ggml_model_path = "https://huggingface.co/CRD716/ggml-vicuna-1.1-quantized/resolve/main/ggml-vicuna-7b-1.1-q4_1.bin"
	filename = "ggml-vicuna-7b-1.1-q4_1.bin"

	download_file(ggml_model_path, filename)

	# n_ctx is max length of prompt
	# n_batch is no. of tokens fed to model at a time
	llm = Llama(model_path=filename, n_ctx=512, n_batch=126)


	def generate_text(prompt="Who is the CEO of Apple?"):
	output = llm(
	prompt,
	max_tokens=256, # max tokens to generate
	temperature=0.1,
	top_p=0.5,
	echo=False, # whether to repeat prompt in output
	stop=["#"],
	)
	output_text = output["choices"][0]["text"].strip()

	# Remove Prompt Echo from Generated Text
	cleaned_output_text = output_text.replace(prompt, "")

	return cleaned_output_text


	description = "Vicuna-7B"

	examples = [
	["What is the capital of France?", "The capital of France is Paris."],
	[
	"Who wrote the novel 'Pride and Prejudice'?",
	"The novel 'Pride and Prejudice' was written by Jane Austen.",
	],
	["What is the square root of 64?", "The square root of 64 is 8."],
	]

	gradio_interface = gr.Interface(
	fn=generate_text,
	inputs=[gr.Textbox(label="Input", lines=2)],
	outputs=[gr.Textbox(label="Result", lines=5)],
	examples=examples,
	title="Vicuna-7B",
	description="Getting output from GGML-based model.",
	)
	gradio_interface.launch()