Spaces:
Runtime error
Runtime error
Wonderplex
commited on
Commit
•
237ffdd
1
Parent(s):
a3f9a2a
added parsing error check (#41)
Browse files
app.py
CHANGED
@@ -16,7 +16,7 @@ from utils import Agent, format_sotopia_prompt, get_starter_prompt, format_bot_m
|
|
16 |
from functools import cache
|
17 |
|
18 |
DEPLOYED = os.getenv("DEPLOYED", "true").lower() == "true"
|
19 |
-
DEFAULT_MODEL_SELECTION = "cmu-lti/sotopia-pi-mistral-7b-BC_SR"
|
20 |
|
21 |
def prepare_sotopia_info():
|
22 |
human_agent = Agent(
|
@@ -59,6 +59,13 @@ def prepare(model_name):
|
|
59 |
)
|
60 |
tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-Instruct-v0.1")
|
61 |
model = PeftModel.from_pretrained(model, model_name).to("cuda")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
62 |
else:
|
63 |
raise RuntimeError(f"Model {model_name} not supported")
|
64 |
return model, tokenizer
|
@@ -107,13 +114,15 @@ def param_accordion(according_visible=True):
|
|
107 |
interactive=True,
|
108 |
label="Max Tokens",
|
109 |
)
|
110 |
-
|
111 |
-
|
112 |
-
|
113 |
-
|
114 |
-
|
|
|
|
|
115 |
)
|
116 |
-
return temperature,
|
117 |
|
118 |
|
119 |
def sotopia_info_accordion(human_agent, machine_agent, scenario, accordion_visible=True):
|
@@ -192,11 +201,20 @@ def chat_tab():
|
|
192 |
text_output = tokenizer.decode(
|
193 |
output_tokens[0], skip_special_tokens=True
|
194 |
)
|
195 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
196 |
|
197 |
with gr.Column():
|
198 |
with gr.Row():
|
199 |
-
temperature,
|
200 |
user_name, bot_name, scenario = sotopia_info_accordion(human_agent, machine_agent, scenario)
|
201 |
|
202 |
instructions = instructions_accordion(instructions)
|
@@ -226,7 +244,7 @@ def chat_tab():
|
|
226 |
user_name,
|
227 |
bot_name,
|
228 |
temperature,
|
229 |
-
|
230 |
max_tokens,
|
231 |
model,
|
232 |
],
|
|
|
16 |
from functools import cache
|
17 |
|
18 |
DEPLOYED = os.getenv("DEPLOYED", "true").lower() == "true"
|
19 |
+
DEFAULT_MODEL_SELECTION = "cmu-lti/sotopia-pi-mistral-7b-BC_SR" # "mistralai/Mistral-7B-Instruct-v0.1"
|
20 |
|
21 |
def prepare_sotopia_info():
|
22 |
human_agent = Agent(
|
|
|
59 |
)
|
60 |
tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-Instruct-v0.1")
|
61 |
model = PeftModel.from_pretrained(model, model_name).to("cuda")
|
62 |
+
elif 'mistralai/Mistral-7B-Instruct-v0.1' in model_name:
|
63 |
+
model = AutoModelForCausalLM.from_pretrained(
|
64 |
+
"mistralai/Mistral-7B-Instruct-v0.1",
|
65 |
+
cache_dir="./.cache",
|
66 |
+
device_map='cuda',
|
67 |
+
)
|
68 |
+
tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-Instruct-v0.1")
|
69 |
else:
|
70 |
raise RuntimeError(f"Model {model_name} not supported")
|
71 |
return model, tokenizer
|
|
|
114 |
interactive=True,
|
115 |
label="Max Tokens",
|
116 |
)
|
117 |
+
top_p = gr.Slider(
|
118 |
+
minimum=1,
|
119 |
+
maximum=3,
|
120 |
+
value=1,
|
121 |
+
interactive=True,
|
122 |
+
visible=True,
|
123 |
+
label="Top p",
|
124 |
)
|
125 |
+
return temperature, top_p, max_tokens, model_name
|
126 |
|
127 |
|
128 |
def sotopia_info_accordion(human_agent, machine_agent, scenario, accordion_visible=True):
|
|
|
201 |
text_output = tokenizer.decode(
|
202 |
output_tokens[0], skip_special_tokens=True
|
203 |
)
|
204 |
+
# import pdb; pdb.set_trace()
|
205 |
+
output = ""
|
206 |
+
for _ in range(5):
|
207 |
+
try:
|
208 |
+
output = format_bot_message(text_output)
|
209 |
+
break
|
210 |
+
except Exception as e:
|
211 |
+
print(e)
|
212 |
+
print("Retrying...")
|
213 |
+
return output
|
214 |
|
215 |
with gr.Column():
|
216 |
with gr.Row():
|
217 |
+
temperature, top_p, max_tokens, model = param_accordion()
|
218 |
user_name, bot_name, scenario = sotopia_info_accordion(human_agent, machine_agent, scenario)
|
219 |
|
220 |
instructions = instructions_accordion(instructions)
|
|
|
244 |
user_name,
|
245 |
bot_name,
|
246 |
temperature,
|
247 |
+
top_p,
|
248 |
max_tokens,
|
249 |
model,
|
250 |
],
|
utils.py
CHANGED
@@ -1,5 +1,6 @@
|
|
1 |
from typing import List, Tuple
|
2 |
import ast
|
|
|
3 |
|
4 |
FORMAT_TEMPLATE = """ Your available action types are
|
5 |
"none action speak non-verbal communication leave".
|
@@ -49,7 +50,12 @@ def truncate_dialogue_history_to_length(dia_his, surpass_num, tokenizer):
|
|
49 |
|
50 |
|
51 |
def format_bot_message(bot_message) -> str:
|
52 |
-
|
|
|
|
|
|
|
|
|
|
|
53 |
match json_response["action_type"]:
|
54 |
case "none":
|
55 |
return 'did nothing'
|
|
|
1 |
from typing import List, Tuple
|
2 |
import ast
|
3 |
+
import re
|
4 |
|
5 |
FORMAT_TEMPLATE = """ Your available action types are
|
6 |
"none action speak non-verbal communication leave".
|
|
|
50 |
|
51 |
|
52 |
def format_bot_message(bot_message) -> str:
|
53 |
+
# import pdb; pdb.set_trace()
|
54 |
+
start_idx, end_idx = bot_message.index("{"), bot_message.index("}")
|
55 |
+
if end_idx == -1:
|
56 |
+
bot_message += "'}"
|
57 |
+
end_idx = len(bot_message)
|
58 |
+
json_response = ast.literal_eval(bot_message[start_idx:end_idx+1])
|
59 |
match json_response["action_type"]:
|
60 |
case "none":
|
61 |
return 'did nothing'
|