File size: 1,927 Bytes
8185d4a 2c45d67 8185d4a 2852c83 04197a7 8185d4a 04197a7 8185d4a 04197a7 8185d4a 2c45d67 8185d4a 2c45d67 2d9e29d 8185d4a 04197a7 b38b5ee 04197a7 8185d4a 2c45d67 04197a7 f27dd35 8185d4a e85c8ba 8185d4a 04197a7 |
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 44 45 46 47 48 49 50 51 52 53 |
import gradio as gr
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch
# Load model and tokenizer
model_name = "Spestly/AwA-1.5B"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.float32, low_cpu_mem_usage=True)
# Set to evaluation mode
model.eval()
def generate_response(message, history):
instruction = (
"You are an LLM called AwA. Aayan Mishra finetunes you. Anthropic does NOT train you. "
"You are a Qwen 2.5 fine-tune. Your purpose is the help the user accomplish their request to the best of your abilities. "
"Below is an instruction that describes a task. Answer it clearly and concisely. Don't overthink answers, but don't underthink them aswell.\n\n"
f"### Instruction:\n{message}\n\n### Response:"
)
inputs = tokenizer(instruction, return_tensors="pt")
with torch.no_grad():
outputs = model.generate(
**inputs,
max_new_tokens=600,
num_return_sequences=1,
temperature=0.7,
top_p=0.9,
do_sample=True
)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
response = response.split("### Response:")[-1].strip()
return response
iface = gr.ChatInterface(
generate_response,
chatbot=gr.Chatbot(height=600, type="messages"),
textbox=gr.Textbox(placeholder="Type your message here...", container=False, scale=7),
title="AwA-1.5B 🔎 - Experimental",
description="Chat with AwA (Answers with Athena). Please note that since AwA is an experimental model, some outputs may not be accurate/expected!",
theme="ocean",
examples=[
"How can CRISPR help us Humans?",
"What are some important ethics in AI",
"What is the capital of New Zealand?",
],
type="messages"
)
iface.launch() |