Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -6,15 +6,39 @@ from threading import Thread
|
|
6 |
import time
|
7 |
import pytz
|
8 |
from datetime import datetime
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
print("Loading model and tokenizer...")
|
10 |
model_name = "large-traversaal/Phi-4-Hindi"
|
11 |
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
12 |
-
model = AutoModelForCausalLM.from_pretrained(
|
|
|
|
|
|
|
|
|
13 |
print("Model and tokenizer loaded successfully!")
|
14 |
-
|
15 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
16 |
start_time = time.time()
|
17 |
-
inputs = tokenizer(
|
18 |
streamer = TextIteratorStreamer(tokenizer, skip_special_tokens=True)
|
19 |
gen_kwargs = {
|
20 |
"input_ids": inputs["input_ids"],
|
@@ -29,16 +53,12 @@ def generate_response(message, temperature, max_new_tokens, top_p):
|
|
29 |
result = []
|
30 |
for text in streamer:
|
31 |
result.append(text)
|
32 |
-
|
33 |
-
if current_output.startswith(message):
|
34 |
-
yield current_output[len(message):]
|
35 |
-
else:
|
36 |
-
yield current_output
|
37 |
end_time = time.time()
|
38 |
time_taken = end_time - start_time
|
39 |
output_text = "".join(result)
|
40 |
-
if output_text
|
41 |
-
output_text = output_text[
|
42 |
print(f"Output: {output_text}")
|
43 |
print(f"Time taken: {time_taken:.2f} seconds")
|
44 |
pst_timezone = pytz.timezone('America/Los_Angeles')
|
@@ -53,6 +73,11 @@ with gr.Blocks() as demo:
|
|
53 |
placeholder="Enter your text here...",
|
54 |
lines=5
|
55 |
)
|
|
|
|
|
|
|
|
|
|
|
56 |
with gr.Row():
|
57 |
with gr.Column():
|
58 |
temperature = gr.Slider(
|
@@ -88,11 +113,11 @@ with gr.Blocks() as demo:
|
|
88 |
)
|
89 |
send_btn.click(
|
90 |
fn=generate_response,
|
91 |
-
inputs=[input_text, temperature, max_new_tokens, top_p],
|
92 |
outputs=output_text
|
93 |
)
|
94 |
clear_btn.click(
|
95 |
-
fn=lambda: ("", ""
|
96 |
inputs=None,
|
97 |
outputs=[input_text, output_text]
|
98 |
)
|
|
|
6 |
import time
|
7 |
import pytz
|
8 |
from datetime import datetime
|
9 |
+
import gradio as gr
|
10 |
+
import torch
|
11 |
+
import time
|
12 |
+
import pytz
|
13 |
+
from datetime import datetime
|
14 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
|
15 |
+
from threading import Thread
|
16 |
print("Loading model and tokenizer...")
|
17 |
model_name = "large-traversaal/Phi-4-Hindi"
|
18 |
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
19 |
+
model = AutoModelForCausalLM.from_pretrained(
|
20 |
+
model_name,
|
21 |
+
torch_dtype=torch.bfloat16,
|
22 |
+
device_map="auto"
|
23 |
+
)
|
24 |
print("Model and tokenizer loaded successfully!")
|
25 |
+
option_mapping = {
|
26 |
+
"translation": "### TRANSLATION ###",
|
27 |
+
"mcq": "### MCQ ###",
|
28 |
+
"nli": "### NLI ###",
|
29 |
+
"summarization": "### SUMMARIZATION ###",
|
30 |
+
"long response": "### LONG RESPONSE ###",
|
31 |
+
"short response": "### SHORT RESPONSE ###",
|
32 |
+
"direct response": "### DIRECT RESPONSE ###",
|
33 |
+
"paraphrase": "### PARAPHRASE ###",
|
34 |
+
"code": "### CODE ###"
|
35 |
+
}
|
36 |
+
def generate_response(message, temperature, max_new_tokens, top_p, task):
|
37 |
+
append_text = option_mapping.get(task, "")
|
38 |
+
prompt = f"INPUT : {message} {append_text} RESPONSE : "
|
39 |
+
print(f"Prompt: {prompt}")
|
40 |
start_time = time.time()
|
41 |
+
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
|
42 |
streamer = TextIteratorStreamer(tokenizer, skip_special_tokens=True)
|
43 |
gen_kwargs = {
|
44 |
"input_ids": inputs["input_ids"],
|
|
|
53 |
result = []
|
54 |
for text in streamer:
|
55 |
result.append(text)
|
56 |
+
yield "".join(result)
|
|
|
|
|
|
|
|
|
57 |
end_time = time.time()
|
58 |
time_taken = end_time - start_time
|
59 |
output_text = "".join(result)
|
60 |
+
if "RESPONSE : " in output_text:
|
61 |
+
output_text = output_text.split("RESPONSE : ", 1)[1].strip()
|
62 |
print(f"Output: {output_text}")
|
63 |
print(f"Time taken: {time_taken:.2f} seconds")
|
64 |
pst_timezone = pytz.timezone('America/Los_Angeles')
|
|
|
73 |
placeholder="Enter your text here...",
|
74 |
lines=5
|
75 |
)
|
76 |
+
task_dropdown = gr.Dropdown(
|
77 |
+
choices=["translation", "mcq", "nli", "summarization", "long response", "short response", "direct response", "paraphrase", "code"],
|
78 |
+
value="long response",
|
79 |
+
label="Task"
|
80 |
+
)
|
81 |
with gr.Row():
|
82 |
with gr.Column():
|
83 |
temperature = gr.Slider(
|
|
|
113 |
)
|
114 |
send_btn.click(
|
115 |
fn=generate_response,
|
116 |
+
inputs=[input_text, temperature, max_new_tokens, top_p, task_dropdown],
|
117 |
outputs=output_text
|
118 |
)
|
119 |
clear_btn.click(
|
120 |
+
fn=lambda: ("", ""),
|
121 |
inputs=None,
|
122 |
outputs=[input_text, output_text]
|
123 |
)
|