Spaces:
Sleeping
Sleeping
File size: 2,363 Bytes
c4ccaf8 73066f5 c4ccaf8 73066f5 c4ccaf8 73066f5 | 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 | import torch
import torch.nn as nn
import numpy as np
import gradio as gr
SEQUENCE_LENGTH = 10
INPUT_SIZE = 1
OUTPUT_SIZE = 1
HIDDEN_UNITS = 128
device = torch.device('cpu')
class Seq2Seq(nn.Module):
def __init__(self, input_size, hidden_size, output_size, seq_len):
super(Seq2Seq, self).__init__()
self.seq_len = seq_len
self.hidden_size = hidden_size
self.encoder_lstm = nn.LSTM(input_size, hidden_size, batch_first=True)
self.decoder_lstm = nn.LSTM(hidden_size, hidden_size, batch_first=True)
self.decoder_linear = nn.Linear(hidden_size, output_size)
def forward(self, x):
_, (hidden, cell) = self.encoder_lstm(x)
context_vector = hidden.permute(1, 0, 2)
decoder_input = context_vector.repeat(1, self.seq_len, 1)
decoder_output, _ = self.decoder_lstm(decoder_input, (hidden, cell))
prediction = self.decoder_linear(decoder_output)
return prediction
model_path = 'seq2seq_model_weights.pth'
model = Seq2Seq(INPUT_SIZE, HIDDEN_UNITS, OUTPUT_SIZE, SEQUENCE_LENGTH).to(device)
model.load_state_dict(torch.load(model_path, map_location=device))
model.eval()
def predict_sequence(input_text):
try:
numbers = [float(n.strip()) for n in input_text.split(',')]
if len(numbers) != SEQUENCE_LENGTH:
return f"Error: Please enter exactly {SEQUENCE_LENGTH} numbers, separated by commas."
input_array = np.array(numbers).reshape(1, SEQUENCE_LENGTH, 1)
input_tensor = torch.from_numpy(input_array).float().to(device)
with torch.no_grad():
prediction_tensor = model(input_tensor)
output_array = prediction_tensor.cpu().numpy().flatten()
output_text = ", ".join([f"{n:.1f}" for n in output_array])
return output_text
except Exception as e:
return f"An error occurred: {str(e)}"
demo = gr.Interface(
fn=predict_sequence,
inputs=gr.Textbox(
label="Input Sequence",
placeholder=f"Enter {SEQUENCE_LENGTH} numbers, e.g., 1, 2, 3, 4, 5, 6, 7, 8, 9, 10"
),
outputs=gr.Textbox(label="Predicted Sequence"),
title="Q11: Seq2Seq Model (n -> n+1)",
description="Enter a sequence of 10 numbers to predict the next sequence.",
allow_flagging="never"
)
demo.launch() |