Spaces:
Runtime error
Runtime error
File size: 2,004 Bytes
899f01e 8fd4e48 899f01e 8fd4e48 899f01e 8fd4e48 |
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 |
import gradio as gr
import transformers
import torch.nn.functional as F
import numpy as np
def generate(model_name="Salesforce/codegen-350M-mono", text="World"):
model = transformers.AutoModelForCausalLM.from_pretrained(model_name)
tokenizer = transformers.AutoTokenizer.from_pretrained(model_name)
input_ids = tokenizer.encode(text, return_tensors='pt')
output = model.generate(input_ids, max_length=100, do_sample=True)
return tokenizer.decode(output[0])
def get_token_likelyhoods(model_name="Salesforce/codegen-350M-mono", text="World"):
# get likelyhoods for each token
model = transformers.AutoModelForCausalLM.from_pretrained(model_name)
tokenizer = transformers.AutoTokenizer.from_pretrained(model_name)
input_ids = tokenizer.encode(text, return_tensors='pt')
out = model(input_ids)
probs = F.softmax(out.logits, dim=-1).squeeze()
output = []
for tok, logits in zip(input_ids.squeeze()[1:], probs):
output.append((
tokenizer.decode(tok),
str(round(logits[tok].item() * 100, 4)) + "%",
# tokenizer.decode(np.argmax(logits.detach()))
))
return output
demo = gr.Interface(
fn=get_token_likelyhoods,
title="Per-token likelyhood GUI based on Giant Language model Test Room",
inputs = [
gr.Textbox(
label="Model name",
lines=1,
value="Salesforce/codegen-350M-mono",
),
gr.Textbox(
label="Text",
lines=3,
value="def first_n_primes(n):\n primes = []\n i = 2\n while len(primes) < n:\n if is_prime(i):\n primes.append(i)\n i += 1\n return",
),
],
outputs = gr.HighlightedText(
label="Diff",
combine_adjacent=True,
).style(color_map={"+": "red", "-": "green"}),
)
if __name__ == "__main__":
demo.launch()
# iface = gr.Interface(fn=generate, inputs=["text", "text"], outputs="text")
# iface.launch()
|