joaogante's picture
joaogante HF staff
typo
a5e42e5
raw
history blame contribute delete
No virus
4.4 kB
import gradio as gr
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
import numpy as np
MODEL_NAME = "google/flan-t5-base"
if __name__ == "__main__":
# Define your model and your tokenizer
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
model = AutoModelForSeq2SeqLM.from_pretrained(MODEL_NAME) # or AutoModelForCausalLM
if tokenizer.pad_token_id is None:
tokenizer.pad_token_id = tokenizer.eos_token_id
model.config.pad_token_id = model.config.eos_token_id
# Define your color-coding labels; if prob > x, then label = y; Sorted in descending probability order!
probs_to_label = [
(0.1, "p >= 10%"),
(0.01, "p >= 1%"),
(1e-20, "p < 1%"),
]
label_to_color = {
"p >= 10%": "green",
"p >= 1%": "yellow",
"p < 1%": "red"
}
def get_tokens_and_labels(prompt):
"""
Given the prompt (text), return a list of tuples (decoded_token, label)
"""
inputs = tokenizer([prompt], return_tensors="pt")
outputs = model.generate(
**inputs, max_new_tokens=50, return_dict_in_generate=True, output_scores=True
)
# Important: don't forget to set `normalize_logits=True` to obtain normalized probabilities (i.e. sum(p) = 1)
transition_scores = model.compute_transition_scores(outputs.sequences, outputs.scores, normalize_logits=True)
transition_proba = np.exp(transition_scores)
# We only have scores for the generated tokens, so pop out the prompt tokens
input_length = 1 if model.config.is_encoder_decoder else inputs.input_ids.shape[1]
generated_ids = outputs.sequences[:, input_length:]
generated_tokens = tokenizer.convert_ids_to_tokens(generated_ids[0])
# Important: you might need to find a tokenization character to replace (e.g. "Δ " for BPE) and get the correct
# spacing into the final output πŸ‘Ό
if model.config.is_encoder_decoder:
highlighted_out = []
else:
input_tokens = tokenizer.convert_ids_to_tokens(inputs.input_ids)
highlighted_out = [(token.replace("▁", " "), None) for token in input_tokens]
# Get the (decoded_token, label) pairs for the generated tokens
for token, proba in zip(generated_tokens, transition_proba[0]):
this_label = None
assert 0. <= proba <= 1.0
for min_proba, label in probs_to_label:
if proba >= min_proba:
this_label = label
break
highlighted_out.append((token.replace("▁", " "), this_label))
return highlighted_out
demo = gr.Blocks()
with demo:
gr.Markdown(
"""
# 🌈 Color-Coded Text Generation 🌈
This is a demo of how you can obtain the probabilities of each generated token, and use them to
color code the model output. Internally, it relies on
[`compute_transition_scores`](https://huggingface.co/docs/transformers/main/en/main_classes/text_generation#transformers.GenerationMixin.compute_transition_scores),
which was added in `transformers` v4.26.0.
⚠️ For instance, with the pre-populated input and its color-coded output, you can see that
`google/flan-t5-base` struggles with arithmetics.
πŸ€— Feel free to clone this demo and modify it to your needs πŸ€—
"""
)
with gr.Row():
with gr.Column():
prompt = gr.Textbox(
label="Prompt",
lines=3,
value=(
"Answer the following question by reasoning step-by-step. The cafeteria had 23 apples. "
"If they used 20 for lunch and bought 6 more, how many apples do they have?"
),
)
button = gr.Button(f"Generate with {MODEL_NAME}")
with gr.Column():
highlighted_text = gr.HighlightedText(
label="Highlighted generation",
combine_adjacent=True,
show_legend=True,
).style(color_map=label_to_color)
button.click(get_tokens_and_labels, inputs=prompt, outputs=highlighted_text)
if __name__ == "__main__":
demo.launch()