Spaces:
Sleeping
Sleeping
File size: 2,230 Bytes
810e1b9 |
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 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 |
import gradio as gr
import time
from transformers import AutoTokenizer, AutoModelForCausalLM, TextGenerationPipeline
MODEL_ID = "akshaynayaks9845/rml-ai-phi1_5-rml-100k"
def load_pipeline():
try:
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
model = AutoModelForCausalLM.from_pretrained(MODEL_ID)
pipe = TextGenerationPipeline(model=model, tokenizer=tokenizer, device=-1)
return pipe
except Exception as e:
return str(e)
pipe_or_err = load_pipeline()
SAMPLES = [
"What is artificial intelligence?",
"Explain machine learning in one sentence.",
"What is quantum computing?",
]
def generate_response(prompt, max_new_tokens=128, temperature=0.2):
start = time.time()
if isinstance(pipe_or_err, str):
return f"Model load error: {pipe_or_err}"
try:
outputs = pipe_or_err(
prompt,
max_new_tokens=int(max_new_tokens),
do_sample=bool(temperature and temperature > 0),
temperature=float(temperature),
top_p=0.9,
repetition_penalty=1.1,
truncation=True,
)
text = outputs[0]["generated_text"]
# Return only continuation if the model echoes the prompt
reply = text[len(prompt):].strip() if text.startswith(prompt) else text
elapsed = int((time.time() - start) * 1000)
return f"{reply}
(⏱️ {elapsed} ms)"
except Exception as e:
return f"Error: {str(e)}"
with gr.Blocks(title="RML-AI Demo") as demo:
gr.Markdown('''
# RML-AI Demo
Ask a question below. The model will respond in GPT-style. This is a lightweight prototype demo.
''')
with gr.Row():
prompt = gr.Textbox(label="Your question", value=SAMPLES[0])
with gr.Row():
max_new = gr.Slider(32, 256, value=128, step=16, label="Max new tokens")
temp = gr.Slider(0.0, 1.0, value=0.2, step=0.1, label="Temperature")
with gr.Row():
btn = gr.Button("Generate")
output = gr.Textbox(label="Answer", lines=8)
with gr.Row():
gr.Examples(SAMPLES, inputs=prompt)
btn.click(generate_response, [prompt, max_new, temp], output)
if __name__ == "__main__":
demo.launch()
|