Spaces:
Runtime error
Runtime error
Merge pull request #28 from Y-IAB/27-instruction
Browse files- app.py +37 -109
- response.py +100 -0
app.py
CHANGED
@@ -3,35 +3,25 @@ It provides a platform for comparing the responses of two LLMs.
|
|
3 |
"""
|
4 |
|
5 |
import enum
|
6 |
-
from random import sample
|
7 |
from uuid import uuid4
|
8 |
|
9 |
import firebase_admin
|
10 |
from firebase_admin import firestore
|
11 |
import gradio as gr
|
12 |
-
from litellm import completion
|
13 |
|
14 |
from leaderboard import build_leaderboard
|
|
|
|
|
15 |
|
16 |
# TODO(#21): Fix auto-reload issue related to the initialization of Firebase.
|
17 |
db_app = firebase_admin.initialize_app()
|
18 |
db = firestore.client()
|
19 |
|
20 |
-
# TODO(#1): Add more models.
|
21 |
-
SUPPORTED_MODELS = [
|
22 |
-
"gpt-4", "gpt-4-0125-preview", "gpt-3.5-turbo", "gemini-pro"
|
23 |
-
]
|
24 |
-
|
25 |
SUPPORTED_TRANSLATION_LANGUAGES = [
|
26 |
"Korean", "English", "Chinese", "Japanese", "Spanish", "French"
|
27 |
]
|
28 |
|
29 |
|
30 |
-
class ResponseType(enum.Enum):
|
31 |
-
SUMMARIZE = "Summarize"
|
32 |
-
TRANSLATE = "Translate"
|
33 |
-
|
34 |
-
|
35 |
class VoteOptions(enum.Enum):
|
36 |
MODEL_A = "Model A is better"
|
37 |
MODEL_B = "Model B is better"
|
@@ -39,107 +29,40 @@ class VoteOptions(enum.Enum):
|
|
39 |
|
40 |
|
41 |
def vote(vote_button, response_a, response_b, model_a_name, model_b_name,
|
42 |
-
user_prompt,
|
43 |
doc_id = uuid4().hex
|
44 |
winner = VoteOptions(vote_button).name.lower()
|
45 |
|
46 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
47 |
doc_ref = db.collection("arena-summarizations").document(doc_id)
|
48 |
-
doc_ref.set(
|
49 |
-
"id": doc_id,
|
50 |
-
"prompt": user_prompt,
|
51 |
-
"model_a": model_a_name,
|
52 |
-
"model_b": model_b_name,
|
53 |
-
"model_a_response": response_a,
|
54 |
-
"model_b_response": response_b,
|
55 |
-
"winner": winner,
|
56 |
-
"timestamp": firestore.SERVER_TIMESTAMP
|
57 |
-
})
|
58 |
return
|
59 |
|
60 |
-
if
|
61 |
doc_ref = db.collection("arena-translations").document(doc_id)
|
62 |
-
|
63 |
-
|
64 |
-
|
65 |
-
"model_a": model_a_name,
|
66 |
-
"model_b": model_b_name,
|
67 |
-
"model_a_response": response_a,
|
68 |
-
"model_b_response": response_b,
|
69 |
-
"source_language": source_lang.lower(),
|
70 |
-
"target_language": target_lang.lower(),
|
71 |
-
"winner": winner,
|
72 |
-
"timestamp": firestore.SERVER_TIMESTAMP
|
73 |
-
})
|
74 |
-
|
75 |
-
|
76 |
-
def response_generator(response: str):
|
77 |
-
for part in response:
|
78 |
-
content = part.choices[0].delta.content
|
79 |
-
if content is None:
|
80 |
-
continue
|
81 |
-
|
82 |
-
# To simulate a stream, we yield each character of the response.
|
83 |
-
for character in content:
|
84 |
-
yield character
|
85 |
-
|
86 |
-
|
87 |
-
def get_responses(user_prompt):
|
88 |
-
models = sample(SUPPORTED_MODELS, 2)
|
89 |
-
|
90 |
-
generators = []
|
91 |
-
for model in models:
|
92 |
-
try:
|
93 |
-
# TODO(#1): Allow user to set configuration.
|
94 |
-
response = completion(model=model,
|
95 |
-
messages=[{
|
96 |
-
"content": user_prompt,
|
97 |
-
"role": "user"
|
98 |
-
}],
|
99 |
-
stream=True)
|
100 |
-
generators.append(response_generator(response))
|
101 |
-
|
102 |
-
# TODO(#1): Narrow down the exception type.
|
103 |
-
except Exception as e: # pylint: disable=broad-except
|
104 |
-
print(f"Error in bot_response: {e}")
|
105 |
-
raise e
|
106 |
-
|
107 |
-
responses = ["", ""]
|
108 |
-
|
109 |
-
# It simulates concurrent response generation from two models.
|
110 |
-
while True:
|
111 |
-
stop = True
|
112 |
-
|
113 |
-
for i in range(len(generators)):
|
114 |
-
try:
|
115 |
-
yielded = next(generators[i])
|
116 |
-
|
117 |
-
if yielded is None:
|
118 |
-
continue
|
119 |
-
|
120 |
-
responses[i] += yielded
|
121 |
-
stop = False
|
122 |
-
|
123 |
-
yield responses + models
|
124 |
-
|
125 |
-
except StopIteration:
|
126 |
-
pass
|
127 |
-
|
128 |
-
# TODO(#1): Narrow down the exception type.
|
129 |
-
except Exception as e: # pylint: disable=broad-except
|
130 |
-
print(f"Error in generator: {e}")
|
131 |
-
raise e
|
132 |
-
|
133 |
-
if stop:
|
134 |
-
break
|
135 |
|
136 |
|
137 |
with gr.Blocks(title="Arena") as app:
|
138 |
with gr.Row():
|
139 |
-
|
140 |
-
[
|
141 |
-
label="
|
142 |
-
info="
|
143 |
|
144 |
source_language = gr.Dropdown(
|
145 |
choices=SUPPORTED_TRANSLATION_LANGUAGES,
|
@@ -154,15 +77,15 @@ with gr.Blocks(title="Arena") as app:
|
|
154 |
interactive=True,
|
155 |
visible=False)
|
156 |
|
157 |
-
def update_language_visibility(
|
158 |
-
visible =
|
159 |
return {
|
160 |
source_language: gr.Dropdown(visible=visible),
|
161 |
target_language: gr.Dropdown(visible=visible)
|
162 |
}
|
163 |
|
164 |
-
|
165 |
-
|
166 |
|
167 |
model_names = [gr.State(None), gr.State(None)]
|
168 |
response_boxes = [gr.State(None), gr.State(None)]
|
@@ -175,7 +98,7 @@ with gr.Blocks(title="Arena") as app:
|
|
175 |
response_boxes[1] = gr.Textbox(label="Model B", interactive=False)
|
176 |
|
177 |
# TODO(#5): Display it only after the user submits the prompt.
|
178 |
-
# TODO(#6): Block voting if the
|
179 |
# TODO(#6): Block voting if the user already voted.
|
180 |
with gr.Row():
|
181 |
option_a = gr.Button(VoteOptions.MODEL_A.value)
|
@@ -188,10 +111,15 @@ with gr.Blocks(title="Arena") as app:
|
|
188 |
model_names[0] = gr.Textbox(label="Model A", interactive=False)
|
189 |
model_names[1] = gr.Textbox(label="Model B", interactive=False)
|
190 |
|
191 |
-
|
|
|
|
|
|
|
|
|
192 |
|
193 |
common_inputs = response_boxes + model_names + [
|
194 |
-
prompt,
|
|
|
195 |
]
|
196 |
option_a.click(vote, [option_a] + common_inputs)
|
197 |
option_b.click(vote, [option_b] + common_inputs)
|
|
|
3 |
"""
|
4 |
|
5 |
import enum
|
|
|
6 |
from uuid import uuid4
|
7 |
|
8 |
import firebase_admin
|
9 |
from firebase_admin import firestore
|
10 |
import gradio as gr
|
|
|
11 |
|
12 |
from leaderboard import build_leaderboard
|
13 |
+
import response
|
14 |
+
from response import get_responses
|
15 |
|
16 |
# TODO(#21): Fix auto-reload issue related to the initialization of Firebase.
|
17 |
db_app = firebase_admin.initialize_app()
|
18 |
db = firestore.client()
|
19 |
|
|
|
|
|
|
|
|
|
|
|
20 |
SUPPORTED_TRANSLATION_LANGUAGES = [
|
21 |
"Korean", "English", "Chinese", "Japanese", "Spanish", "French"
|
22 |
]
|
23 |
|
24 |
|
|
|
|
|
|
|
|
|
|
|
25 |
class VoteOptions(enum.Enum):
|
26 |
MODEL_A = "Model A is better"
|
27 |
MODEL_B = "Model B is better"
|
|
|
29 |
|
30 |
|
31 |
def vote(vote_button, response_a, response_b, model_a_name, model_b_name,
|
32 |
+
user_prompt, instruction, category, source_lang, target_lang):
|
33 |
doc_id = uuid4().hex
|
34 |
winner = VoteOptions(vote_button).name.lower()
|
35 |
|
36 |
+
doc = {
|
37 |
+
"id": doc_id,
|
38 |
+
"prompt": user_prompt,
|
39 |
+
"instruction": instruction,
|
40 |
+
"model_a": model_a_name,
|
41 |
+
"model_b": model_b_name,
|
42 |
+
"model_a_response": response_a,
|
43 |
+
"model_b_response": response_b,
|
44 |
+
"winner": winner,
|
45 |
+
"timestamp": firestore.SERVER_TIMESTAMP
|
46 |
+
}
|
47 |
+
|
48 |
+
if category == response.Category.SUMMARIZE.value:
|
49 |
doc_ref = db.collection("arena-summarizations").document(doc_id)
|
50 |
+
doc_ref.set(doc)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
51 |
return
|
52 |
|
53 |
+
if category == response.Category.TRANSLATE.value:
|
54 |
doc_ref = db.collection("arena-translations").document(doc_id)
|
55 |
+
doc["source_lang"] = source_lang.lower()
|
56 |
+
doc["target_lang"] = target_lang.lower()
|
57 |
+
doc_ref.set(doc)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
58 |
|
59 |
|
60 |
with gr.Blocks(title="Arena") as app:
|
61 |
with gr.Row():
|
62 |
+
category_radio = gr.Radio(
|
63 |
+
[category.value for category in response.Category],
|
64 |
+
label="Category",
|
65 |
+
info="The chosen category determines the instruction sent to the LLMs.")
|
66 |
|
67 |
source_language = gr.Dropdown(
|
68 |
choices=SUPPORTED_TRANSLATION_LANGUAGES,
|
|
|
77 |
interactive=True,
|
78 |
visible=False)
|
79 |
|
80 |
+
def update_language_visibility(category):
|
81 |
+
visible = category == response.Category.TRANSLATE.value
|
82 |
return {
|
83 |
source_language: gr.Dropdown(visible=visible),
|
84 |
target_language: gr.Dropdown(visible=visible)
|
85 |
}
|
86 |
|
87 |
+
category_radio.change(update_language_visibility, category_radio,
|
88 |
+
[source_language, target_language])
|
89 |
|
90 |
model_names = [gr.State(None), gr.State(None)]
|
91 |
response_boxes = [gr.State(None), gr.State(None)]
|
|
|
98 |
response_boxes[1] = gr.Textbox(label="Model B", interactive=False)
|
99 |
|
100 |
# TODO(#5): Display it only after the user submits the prompt.
|
101 |
+
# TODO(#6): Block voting if the category is not set.
|
102 |
# TODO(#6): Block voting if the user already voted.
|
103 |
with gr.Row():
|
104 |
option_a = gr.Button(VoteOptions.MODEL_A.value)
|
|
|
111 |
model_names[0] = gr.Textbox(label="Model A", interactive=False)
|
112 |
model_names[1] = gr.Textbox(label="Model B", interactive=False)
|
113 |
|
114 |
+
instruction_state = gr.State("")
|
115 |
+
|
116 |
+
submit.click(get_responses,
|
117 |
+
[prompt, category_radio, source_language, target_language],
|
118 |
+
response_boxes + model_names + [instruction_state])
|
119 |
|
120 |
common_inputs = response_boxes + model_names + [
|
121 |
+
prompt, instruction_state, category_radio, source_language,
|
122 |
+
target_language
|
123 |
]
|
124 |
option_a.click(vote, [option_a] + common_inputs)
|
125 |
option_b.click(vote, [option_b] + common_inputs)
|
response.py
ADDED
@@ -0,0 +1,100 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
"""
|
2 |
+
This module contains functions for generating responses using LLMs.
|
3 |
+
"""
|
4 |
+
|
5 |
+
import enum
|
6 |
+
from random import sample
|
7 |
+
|
8 |
+
import gradio as gr
|
9 |
+
from litellm import completion
|
10 |
+
|
11 |
+
# TODO(#1): Add more models.
|
12 |
+
SUPPORTED_MODELS = [
|
13 |
+
"gpt-4", "gpt-4-0125-preview", "gpt-3.5-turbo", "gemini-pro"
|
14 |
+
]
|
15 |
+
|
16 |
+
|
17 |
+
class Category(enum.Enum):
|
18 |
+
SUMMARIZE = "Summarize"
|
19 |
+
TRANSLATE = "Translate"
|
20 |
+
|
21 |
+
|
22 |
+
# TODO(#31): Let the model builders set the instruction.
|
23 |
+
def get_instruction(category, source_lang, target_lang):
|
24 |
+
if category == Category.SUMMARIZE.value:
|
25 |
+
return "Summarize the following text in its original language."
|
26 |
+
if category == Category.TRANSLATE.value:
|
27 |
+
return f"Translate the following text from {source_lang} to {target_lang}."
|
28 |
+
|
29 |
+
|
30 |
+
def response_generator(response: str):
|
31 |
+
for part in response:
|
32 |
+
content = part.choices[0].delta.content
|
33 |
+
if content is None:
|
34 |
+
continue
|
35 |
+
|
36 |
+
# To simulate a stream, we yield each character of the response.
|
37 |
+
for character in content:
|
38 |
+
yield character
|
39 |
+
|
40 |
+
|
41 |
+
# TODO(#29): Return results simultaneously to prevent bias from generation speed.
|
42 |
+
def get_responses(user_prompt, category, source_lang, target_lang):
|
43 |
+
if not category:
|
44 |
+
raise gr.Error("Please select a category.")
|
45 |
+
|
46 |
+
if category == Category.TRANSLATE.value and (not source_lang or
|
47 |
+
not target_lang):
|
48 |
+
raise gr.Error("Please select source and target languages.")
|
49 |
+
|
50 |
+
models = sample(SUPPORTED_MODELS, 2)
|
51 |
+
instruction = get_instruction(category, source_lang, target_lang)
|
52 |
+
|
53 |
+
generators = []
|
54 |
+
for model in models:
|
55 |
+
try:
|
56 |
+
# TODO(#1): Allow user to set configuration.
|
57 |
+
response = completion(model=model,
|
58 |
+
messages=[{
|
59 |
+
"content": instruction,
|
60 |
+
"role": "system"
|
61 |
+
}, {
|
62 |
+
"content": user_prompt,
|
63 |
+
"role": "user"
|
64 |
+
}],
|
65 |
+
stream=True)
|
66 |
+
generators.append(response_generator(response))
|
67 |
+
|
68 |
+
# TODO(#1): Narrow down the exception type.
|
69 |
+
except Exception as e: # pylint: disable=broad-except
|
70 |
+
print(f"Error in bot_response: {e}")
|
71 |
+
raise e
|
72 |
+
|
73 |
+
responses = ["", ""]
|
74 |
+
|
75 |
+
# It simulates concurrent response generation from two models.
|
76 |
+
while True:
|
77 |
+
stop = True
|
78 |
+
|
79 |
+
for i in range(len(generators)):
|
80 |
+
try:
|
81 |
+
yielded = next(generators[i])
|
82 |
+
|
83 |
+
if yielded is None:
|
84 |
+
continue
|
85 |
+
|
86 |
+
responses[i] += yielded
|
87 |
+
stop = False
|
88 |
+
|
89 |
+
yield responses + models + [instruction]
|
90 |
+
|
91 |
+
except StopIteration:
|
92 |
+
pass
|
93 |
+
|
94 |
+
# TODO(#1): Narrow down the exception type.
|
95 |
+
except Exception as e: # pylint: disable=broad-except
|
96 |
+
print(f"Error in generator: {e}")
|
97 |
+
raise e
|
98 |
+
|
99 |
+
if stop:
|
100 |
+
break
|