Delete inference.py
Browse files- inference.py +0 -28
inference.py
DELETED
@@ -1,28 +0,0 @@
|
|
1 |
-
import gradio as gr
|
2 |
-
from transformers import AutoTokenizer, TFBlenderbotForConditionalGeneration
|
3 |
-
import tensorflow as tf
|
4 |
-
model_name = "WICKED4950/Irisonego5"
|
5 |
-
strategy = tf.distribute.MirroredStrategy()
|
6 |
-
tf.config.optimizer.set_jit(True) # Enable XLA
|
7 |
-
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
8 |
-
with strategy.scope():
|
9 |
-
model = TFBlenderbotForConditionalGeneration.from_pretrained(model_name)
|
10 |
-
|
11 |
-
# Define the chatbot function
|
12 |
-
def predict(user_input):
|
13 |
-
# Tokenize input text
|
14 |
-
inputs = tokenizer(user_input, return_tensors="tf", padding=True, truncation=True)
|
15 |
-
|
16 |
-
# Generate the response using the model
|
17 |
-
response_ids = model.generate(
|
18 |
-
inputs['input_ids'],
|
19 |
-
max_length=128, # Set max length of response
|
20 |
-
do_sample=True, # Sampling for variability
|
21 |
-
top_k=15, # Consider top 50 tokens
|
22 |
-
top_p=0.95, # Nucleus sampling
|
23 |
-
temperature=0.8 # Adjusts creativity of response
|
24 |
-
)
|
25 |
-
|
26 |
-
# Decode the response
|
27 |
-
response = tokenizer.decode(response_id[0], skip_special_tokens=True)
|
28 |
-
return response
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|