Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
@@ -2,22 +2,23 @@ import torch
|
|
2 |
from PIL import Image
|
3 |
import gradio as gr
|
4 |
import spaces
|
5 |
-
from transformers import AutoModelForCausalLM,
|
6 |
import os
|
7 |
from threading import Thread
|
8 |
|
9 |
|
10 |
HF_TOKEN = os.environ.get("HF_TOKEN", None)
|
11 |
-
MODEL_ID = "
|
12 |
MODELS = os.environ.get("MODELS")
|
13 |
MODEL_NAME = MODELS.split("/")[-1]
|
|
|
14 |
|
15 |
TITLE = "<h1><center>Qwen2-Chatbox</center></h1>"
|
16 |
|
17 |
DESCRIPTION = f"""
|
18 |
<h3>MODEL: <a href="https://hf.co/{MODELS}">{MODEL_NAME}</a></h3>
|
19 |
<center>
|
20 |
-
<p>
|
21 |
<br>
|
22 |
Feel free to test without log.
|
23 |
</p>
|
@@ -37,13 +38,15 @@ h3 {
|
|
37 |
"""
|
38 |
|
39 |
model = AutoModelForCausalLM.from_pretrained(
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
)
|
44 |
-
tokenizer =
|
|
|
|
|
45 |
|
46 |
-
@spaces.GPU
|
47 |
def stream_chat(message: str, history: list, temperature: float, max_new_tokens: int, top_p: float, top_k: int, penalty: float):
|
48 |
print(f'message is - {message}')
|
49 |
print(f'history is - {history}')
|
@@ -54,13 +57,16 @@ def stream_chat(message: str, history: list, temperature: float, max_new_tokens:
|
|
54 |
|
55 |
print(f"Conversation is -\n{conversation}")
|
56 |
|
57 |
-
input_ids = tokenizer.apply_chat_template(conversation,
|
58 |
-
|
|
|
|
|
|
|
59 |
|
60 |
-
streamer = TextIteratorStreamer(tokenizer, timeout=
|
61 |
|
62 |
generate_kwargs = dict(
|
63 |
-
|
64 |
streamer=streamer,
|
65 |
top_k=top_k,
|
66 |
top_p=top_p,
|
@@ -68,7 +74,8 @@ def stream_chat(message: str, history: list, temperature: float, max_new_tokens:
|
|
68 |
max_new_tokens=max_new_tokens,
|
69 |
do_sample=True,
|
70 |
temperature=temperature,
|
71 |
-
|
|
|
72 |
)
|
73 |
|
74 |
thread = Thread(target=model.generate, kwargs=generate_kwargs)
|
@@ -81,9 +88,9 @@ def stream_chat(message: str, history: list, temperature: float, max_new_tokens:
|
|
81 |
|
82 |
|
83 |
|
84 |
-
chatbot = gr.Chatbot(height=
|
85 |
|
86 |
-
with gr.Blocks(css=CSS) as demo:
|
87 |
gr.HTML(TITLE)
|
88 |
gr.HTML(DESCRIPTION)
|
89 |
gr.DuplicateButton(value="Duplicate Space for private use", elem_classes="duplicate-button")
|
@@ -103,7 +110,7 @@ with gr.Blocks(css=CSS) as demo:
|
|
103 |
),
|
104 |
gr.Slider(
|
105 |
minimum=128,
|
106 |
-
maximum=
|
107 |
step=1,
|
108 |
value=1024,
|
109 |
label="Max new tokens",
|
|
|
2 |
from PIL import Image
|
3 |
import gradio as gr
|
4 |
import spaces
|
5 |
+
from transformers import AutoModelForCausalLM, GemmaTokenizerFast, TextIteratorStreamer,BitsAndBytesConfig
|
6 |
import os
|
7 |
from threading import Thread
|
8 |
|
9 |
|
10 |
HF_TOKEN = os.environ.get("HF_TOKEN", None)
|
11 |
+
MODEL_ID = "google/gemma-2-27b-it"
|
12 |
MODELS = os.environ.get("MODELS")
|
13 |
MODEL_NAME = MODELS.split("/")[-1]
|
14 |
+
MAX_INPUT_TOKEN_LENGTH = int(os.environ.get("MAX_INPUT_TOKEN_LENGTH", "4096"))
|
15 |
|
16 |
TITLE = "<h1><center>Qwen2-Chatbox</center></h1>"
|
17 |
|
18 |
DESCRIPTION = f"""
|
19 |
<h3>MODEL: <a href="https://hf.co/{MODELS}">{MODEL_NAME}</a></h3>
|
20 |
<center>
|
21 |
+
<p>Gemma is the large language model built by Google.
|
22 |
<br>
|
23 |
Feel free to test without log.
|
24 |
</p>
|
|
|
38 |
"""
|
39 |
|
40 |
model = AutoModelForCausalLM.from_pretrained(
|
41 |
+
MODELS,
|
42 |
+
device_map="auto",
|
43 |
+
quantization_config=BitsAndBytesConfig(load_in_4bit=True)
|
44 |
)
|
45 |
+
tokenizer = GemmaTokenizerFast.from_pretrained(MODELS)
|
46 |
+
model.config.sliding_window = 4096
|
47 |
+
model.eval()
|
48 |
|
49 |
+
@spaces.GPU(duration=90)
|
50 |
def stream_chat(message: str, history: list, temperature: float, max_new_tokens: int, top_p: float, top_k: int, penalty: float):
|
51 |
print(f'message is - {message}')
|
52 |
print(f'history is - {history}')
|
|
|
57 |
|
58 |
print(f"Conversation is -\n{conversation}")
|
59 |
|
60 |
+
input_ids = tokenizer.apply_chat_template(conversation, add_generation_prompt=True, return_tensors="pt")
|
61 |
+
if input_ids.shape[1] > MAX_INPUT_TOKEN_LENGTH:
|
62 |
+
input_ids = input_ids[:, -MAX_INPUT_TOKEN_LENGTH:]
|
63 |
+
gr.Warning(f"Trimmed input from conversation as it was longer than {MAX_INPUT_TOKEN_LENGTH} tokens.")
|
64 |
+
input_ids = input_ids.to(0)
|
65 |
|
66 |
+
streamer = TextIteratorStreamer(tokenizer, timeout=20.0, skip_prompt=True, skip_special_tokens=True)
|
67 |
|
68 |
generate_kwargs = dict(
|
69 |
+
{"input_ids": input_ids},
|
70 |
streamer=streamer,
|
71 |
top_k=top_k,
|
72 |
top_p=top_p,
|
|
|
74 |
max_new_tokens=max_new_tokens,
|
75 |
do_sample=True,
|
76 |
temperature=temperature,
|
77 |
+
num_beams=1,
|
78 |
+
repetition_penalty=repetition_penalty,
|
79 |
)
|
80 |
|
81 |
thread = Thread(target=model.generate, kwargs=generate_kwargs)
|
|
|
88 |
|
89 |
|
90 |
|
91 |
+
chatbot = gr.Chatbot(height=600)
|
92 |
|
93 |
+
with gr.Blocks(css=CSS, theme="soft") as demo:
|
94 |
gr.HTML(TITLE)
|
95 |
gr.HTML(DESCRIPTION)
|
96 |
gr.DuplicateButton(value="Duplicate Space for private use", elem_classes="duplicate-button")
|
|
|
110 |
),
|
111 |
gr.Slider(
|
112 |
minimum=128,
|
113 |
+
maximum=2048,
|
114 |
step=1,
|
115 |
value=1024,
|
116 |
label="Max new tokens",
|