Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
@@ -1,6 +1,7 @@
|
|
1 |
import spaces
|
2 |
-
from transformers import AutoModelForCausalLM, AutoTokenizer
|
3 |
import gradio as gr
|
|
|
4 |
|
5 |
checkpoint = "WillHeld/soft-raccoon"
|
6 |
device = "cuda"
|
@@ -13,20 +14,28 @@ def predict(message, history, temperature, top_p):
|
|
13 |
input_text = tokenizer.apply_chat_template(history, tokenize=False, add_generation_prompt=True)
|
14 |
inputs = tokenizer.encode(input_text, return_tensors="pt").to(device)
|
15 |
|
16 |
-
|
|
|
17 |
|
18 |
-
#
|
19 |
-
|
20 |
-
inputs,
|
21 |
-
max_new_tokens
|
22 |
-
temperature
|
23 |
-
top_p
|
24 |
-
do_sample
|
25 |
-
streamer
|
26 |
-
|
|
|
|
|
|
|
|
|
27 |
|
28 |
-
#
|
29 |
-
|
|
|
|
|
|
|
30 |
|
31 |
with gr.Blocks() as demo:
|
32 |
chatbot = gr.ChatInterface(
|
|
|
1 |
import spaces
|
2 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
|
3 |
import gradio as gr
|
4 |
+
from threading import Thread
|
5 |
|
6 |
checkpoint = "WillHeld/soft-raccoon"
|
7 |
device = "cuda"
|
|
|
14 |
input_text = tokenizer.apply_chat_template(history, tokenize=False, add_generation_prompt=True)
|
15 |
inputs = tokenizer.encode(input_text, return_tensors="pt").to(device)
|
16 |
|
17 |
+
# Create a streamer
|
18 |
+
streamer = TextIteratorStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
|
19 |
|
20 |
+
# Set up generation parameters
|
21 |
+
generation_kwargs = {
|
22 |
+
"input_ids": inputs,
|
23 |
+
"max_new_tokens": 1024,
|
24 |
+
"temperature": float(temperature),
|
25 |
+
"top_p": float(top_p),
|
26 |
+
"do_sample": True,
|
27 |
+
"streamer": streamer,
|
28 |
+
}
|
29 |
+
|
30 |
+
# Run generation in a separate thread
|
31 |
+
thread = Thread(target=model.generate, kwargs=generation_kwargs)
|
32 |
+
thread.start()
|
33 |
|
34 |
+
# Yield from the streamer as tokens are generated
|
35 |
+
partial_text = ""
|
36 |
+
for new_text in streamer:
|
37 |
+
partial_text += new_text
|
38 |
+
yield partial_text
|
39 |
|
40 |
with gr.Blocks() as demo:
|
41 |
chatbot = gr.ChatInterface(
|