Spaces:
Runtime error
Runtime error
File size: 1,331 Bytes
f474fd0 0356490 f474fd0 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 |
import torch
from peft import PeftModel
import transformers
import gradio as gr
from transformers import GenerationConfig, LlamaForCausalLM, LlamaTokenizer
from transformers import Trainer
BASE_MODEL = "TheBloke/stable-vicuna-13B-HF"
model = LlamaForCausalLM.from_pretrained(BASE_MODEL, load_in_8bit=True, torch_dtype=torch.float16, device_map="auto", offload_folder="./cache")
tokenizer = LlamaTokenizer.from_pretrained(BASE_MODEL)
tokenizer.pad_token_id = 0
tokenizer.padding_side = "left"
def format_prompt(prompt: str) -> str:
return f"### Human: {prompt}\n### Assistant:"
generation_config = GenerationConfig(
max_new_tokens=128,
temperature=0.2,
repetition_penalty=1.0,
)
def generate_text(prompt: str):
formatted_prompt = format_prompt(prompt)
inputs = tokenizer(
formatted_prompt,
padding=False,
add_special_tokens=False,
return_tensors="pt"
).to(model.device)
with torch.inference_mode():
tokens = model.generate(**inputs, generation_config=generation_config)
response = tokenizer.decode(tokens[0], skip_special_tokens=True)
assistant_index = response.find("### Assistant:") + len("### Assistant:")
return response[assistant_index:].strip()
iface = gr.Interface(fn=generate_text, inputs="text", outputs="text")
iface.launch()
|