Spaces:
Paused
Paused
fix: add system prompt
Browse files
app.py
CHANGED
@@ -1,25 +1,30 @@
|
|
1 |
import os
|
|
|
2 |
from string import Template
|
3 |
from threading import Thread
|
4 |
|
5 |
import torch
|
6 |
import gradio as gr
|
7 |
-
from transformers import AutoTokenizer, AutoModelForCausalLM, TextIteratorStreamer
|
8 |
|
9 |
|
10 |
auth_token = os.environ.get("HUGGINGFACE_TOKEN")
|
11 |
tokenizer = AutoTokenizer.from_pretrained(
|
12 |
-
"CarperAI/vicuna-13b-
|
13 |
use_auth_token=auth_token if auth_token else True,
|
14 |
)
|
15 |
model = AutoModelForCausalLM.from_pretrained(
|
16 |
-
"CarperAI/vicuna-13b-
|
|
|
|
|
|
|
17 |
use_auth_token=auth_token if auth_token else True,
|
18 |
)
|
|
|
19 |
|
20 |
|
21 |
max_context_length = model.config.max_position_embeddings
|
22 |
-
max_new_tokens =
|
23 |
|
24 |
|
25 |
prompt_template = Template("""\
|
@@ -28,10 +33,15 @@ prompt_template = Template("""\
|
|
28 |
""")
|
29 |
|
30 |
|
|
|
|
|
|
|
|
|
|
|
31 |
def bot(history):
|
32 |
-
# print(f"History:\n`{history}`")
|
33 |
history = history or []
|
34 |
-
|
|
|
35 |
prompt_history = []
|
36 |
for human, bot in history:
|
37 |
if bot is not None:
|
@@ -42,54 +52,53 @@ def bot(history):
|
|
42 |
human=human, bot=bot if bot is not None else "")
|
43 |
)
|
44 |
|
45 |
-
|
46 |
-
|
47 |
-
|
|
|
|
|
48 |
|
49 |
-
#
|
50 |
-
#
|
51 |
-
|
52 |
-
inputs = {
|
53 |
-
|
|
|
|
|
|
|
54 |
if inputs.get("token_type_ids", None) is not None:
|
55 |
inputs.pop("token_type_ids")
|
56 |
-
|
57 |
streamer = TextIteratorStreamer(
|
58 |
tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True
|
59 |
)
|
60 |
-
|
61 |
-
# Generate the response
|
62 |
generate_kwargs = dict(
|
63 |
inputs,
|
64 |
streamer=streamer,
|
65 |
max_new_tokens=max_new_tokens,
|
66 |
do_sample=True,
|
|
|
67 |
temperature=1.0,
|
68 |
-
top_p=0.9999,
|
69 |
)
|
70 |
-
|
71 |
-
# print(f"Generating with kwargs: {generate_kwargs}")
|
72 |
thread = Thread(target=model.generate, kwargs=generate_kwargs)
|
73 |
thread.start()
|
74 |
|
75 |
partial_text = ""
|
76 |
for new_text in streamer:
|
77 |
-
# Process out the prompt separator
|
78 |
new_text = new_text.replace("<br>", "\n")
|
79 |
-
# print(f"New text: `{new_text}`")
|
80 |
if "###" in new_text:
|
81 |
new_text = new_text.split("###")[0]
|
82 |
partial_text += new_text.strip()
|
83 |
history[-1][1] = partial_text
|
84 |
break
|
85 |
else:
|
86 |
-
# Filter empty trailing
|
87 |
-
if new_text
|
88 |
new_text = new_text.strip()
|
89 |
partial_text += new_text
|
90 |
history[-1][1] = partial_text
|
91 |
yield history
|
92 |
-
|
93 |
return partial_text
|
94 |
|
95 |
|
@@ -98,28 +107,32 @@ def user(user_message, history):
|
|
98 |
|
99 |
|
100 |
with gr.Blocks() as demo:
|
101 |
-
gr.Markdown("
|
102 |
-
gr.HTML("<a href='https://huggingface.co/
|
103 |
-
gr.HTML('''<center><a href="https://huggingface.co/spaces/
|
104 |
|
105 |
-
chatbot = gr.Chatbot([], elem_id="chatbot").style(height=
|
106 |
state = gr.State([])
|
107 |
with gr.Row():
|
108 |
with gr.Column():
|
109 |
-
msg = gr.Textbox(
|
110 |
-
|
|
|
|
|
|
|
111 |
with gr.Column():
|
112 |
with gr.Row():
|
113 |
-
submit = gr.Button("
|
114 |
stop = gr.Button("Stop")
|
115 |
-
clear = gr.Button("Clear")
|
116 |
-
|
117 |
-
|
118 |
-
|
119 |
-
|
120 |
-
|
121 |
-
|
122 |
-
|
|
|
123 |
|
124 |
demo.queue(max_size=32, concurrency_count=2)
|
125 |
-
demo.launch()
|
|
|
1 |
import os
|
2 |
+
import gc
|
3 |
from string import Template
|
4 |
from threading import Thread
|
5 |
|
6 |
import torch
|
7 |
import gradio as gr
|
8 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM, BatchEncoding, TextIteratorStreamer
|
9 |
|
10 |
|
11 |
auth_token = os.environ.get("HUGGINGFACE_TOKEN")
|
12 |
tokenizer = AutoTokenizer.from_pretrained(
|
13 |
+
"CarperAI/stable-vicuna-13b-fp16",
|
14 |
use_auth_token=auth_token if auth_token else True,
|
15 |
)
|
16 |
model = AutoModelForCausalLM.from_pretrained(
|
17 |
+
"CarperAI/stable-vicuna-13b-fp16",
|
18 |
+
torch_dtype=torch.float16,
|
19 |
+
low_cpu_mem_usage=True,
|
20 |
+
device_map="auto",
|
21 |
use_auth_token=auth_token if auth_token else True,
|
22 |
)
|
23 |
+
model.eval()
|
24 |
|
25 |
|
26 |
max_context_length = model.config.max_position_embeddings
|
27 |
+
max_new_tokens = 768
|
28 |
|
29 |
|
30 |
prompt_template = Template("""\
|
|
|
33 |
""")
|
34 |
|
35 |
|
36 |
+
system_prompt = "### Assistant: I am StableVicuna, a large language model created by Stability AI. I am here to chat!"
|
37 |
+
system_prompt_tokens = tokenizer([f"{system_prompt}\n\n"], return_tensors="pt")
|
38 |
+
max_sys_tokens = system_prompt_tokens['input_ids'].size(-1)
|
39 |
+
|
40 |
+
|
41 |
def bot(history):
|
|
|
42 |
history = history or []
|
43 |
+
|
44 |
+
# Inject prompt formatting into the history
|
45 |
prompt_history = []
|
46 |
for human, bot in history:
|
47 |
if bot is not None:
|
|
|
52 |
human=human, bot=bot if bot is not None else "")
|
53 |
)
|
54 |
|
55 |
+
msg_tokens = tokenizer(
|
56 |
+
"\n\n".join(prompt_history).strip(),
|
57 |
+
return_tensors="pt",
|
58 |
+
add_special_tokens=False # Use <BOS> from the system prompt
|
59 |
+
)
|
60 |
|
61 |
+
# Take only the most recent context up to the max context length and prepend the
|
62 |
+
# system prompt with the messages
|
63 |
+
max_tokens = -max_context_length + max_new_tokens + max_sys_tokens
|
64 |
+
inputs = BatchEncoding({
|
65 |
+
k: torch.concat([system_prompt_tokens[k], msg_tokens[k][:, max_tokens:]], dim=-1)
|
66 |
+
for k in msg_tokens
|
67 |
+
}).to('cuda')
|
68 |
+
# Remove `token_type_ids` b/c it's not yet supported for LLaMA `transformers` models
|
69 |
if inputs.get("token_type_ids", None) is not None:
|
70 |
inputs.pop("token_type_ids")
|
71 |
+
|
72 |
streamer = TextIteratorStreamer(
|
73 |
tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True
|
74 |
)
|
|
|
|
|
75 |
generate_kwargs = dict(
|
76 |
inputs,
|
77 |
streamer=streamer,
|
78 |
max_new_tokens=max_new_tokens,
|
79 |
do_sample=True,
|
80 |
+
top_p=1.0,
|
81 |
temperature=1.0,
|
|
|
82 |
)
|
|
|
|
|
83 |
thread = Thread(target=model.generate, kwargs=generate_kwargs)
|
84 |
thread.start()
|
85 |
|
86 |
partial_text = ""
|
87 |
for new_text in streamer:
|
88 |
+
# Process out the prompt separator
|
89 |
new_text = new_text.replace("<br>", "\n")
|
|
|
90 |
if "###" in new_text:
|
91 |
new_text = new_text.split("###")[0]
|
92 |
partial_text += new_text.strip()
|
93 |
history[-1][1] = partial_text
|
94 |
break
|
95 |
else:
|
96 |
+
# Filter empty trailing new lines
|
97 |
+
if new_text == "\n":
|
98 |
new_text = new_text.strip()
|
99 |
partial_text += new_text
|
100 |
history[-1][1] = partial_text
|
101 |
yield history
|
|
|
102 |
return partial_text
|
103 |
|
104 |
|
|
|
107 |
|
108 |
|
109 |
with gr.Blocks() as demo:
|
110 |
+
gr.Markdown("StableVicuna by Stability AI")
|
111 |
+
gr.HTML("<a href='https://huggingface.co/stabilityai/stable-vicuna-13b-delta'><code>stabilityai/stable-vicuna-13b-delta</a>")
|
112 |
+
gr.HTML('''<center><a href="https://huggingface.co/spaces/stabilityai/stable-vicuna?duplicate=true"><img src="https://bit.ly/3gLdBN6" alt="Duplicate Space"></a>Duplicate the Space to skip the queue and run in a private space</center>''')
|
113 |
|
114 |
+
chatbot = gr.Chatbot([], elem_id="chatbot").style(height=500)
|
115 |
state = gr.State([])
|
116 |
with gr.Row():
|
117 |
with gr.Column():
|
118 |
+
msg = gr.Textbox(
|
119 |
+
label="Send a message",
|
120 |
+
placeholder="Send a message",
|
121 |
+
show_label=False
|
122 |
+
).style(container=False)
|
123 |
with gr.Column():
|
124 |
with gr.Row():
|
125 |
+
submit = gr.Button("Send")
|
126 |
stop = gr.Button("Stop")
|
127 |
+
clear = gr.Button("Clear History")
|
128 |
+
|
129 |
+
submit_event = msg.submit(user, inputs=[msg, chatbot], outputs=[msg, chatbot], queue=False).then(
|
130 |
+
fn=bot, inputs=[chatbot], outputs=[chatbot], queue=True)
|
131 |
+
submit_click_event = submit.click(user, inputs=[msg, chatbot], outputs=[msg, chatbot], queue=False).then(
|
132 |
+
fn=bot, inputs=[chatbot], outputs=[chatbot], queue=True)
|
133 |
+
|
134 |
+
stop.click(fn=None, inputs=None, outputs=None, cancels=[submit_event, submit_click_event], queue=False)
|
135 |
+
clear.click(lambda: None, None, [chatbot], queue=True)
|
136 |
|
137 |
demo.queue(max_size=32, concurrency_count=2)
|
138 |
+
demo.launch(share=True)
|