Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -3,9 +3,51 @@ import transformers
|
|
3 |
import torch
|
4 |
from peft import PeftModel
|
5 |
import os
|
|
|
|
|
|
|
|
|
6 |
|
|
|
|
|
|
|
|
|
7 |
HF_TOKEN = os.environ.get("HF_TOKEN")
|
8 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
model_id = "JerniganLab/interviews-and-qa"
|
10 |
base_model = "meta-llama/Meta-Llama-3-8B-Instruct"
|
11 |
|
@@ -22,6 +64,17 @@ pipeline = transformers.pipeline(
|
|
22 |
|
23 |
pipeline.model = PeftModel.from_pretrained(llama_model, model_id)
|
24 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
25 |
# def chat_function(message, history, system_prompt, max_new_tokens, temperature):
|
26 |
# messages = [{"role":"system","content":system_prompt},
|
27 |
# {"role":"user", "content":message}]
|
@@ -59,6 +112,7 @@ def chat_function(message, history, max_new_tokens, temperature):
|
|
59 |
do_sample = True,
|
60 |
temperature = temperature + 0.1,
|
61 |
top_p = 0.9,)
|
|
|
62 |
return outputs[0]["generated_text"][len(prompt):]
|
63 |
|
64 |
"""
|
@@ -82,10 +136,12 @@ demo = gr.ChatInterface(
|
|
82 |
additional_inputs=[
|
83 |
gr.Slider(100,4000, label="Max New Tokens"),
|
84 |
gr.Slider(0,1, label="Temperature")
|
85 |
-
]
|
|
|
|
|
86 |
)
|
87 |
|
88 |
|
89 |
|
90 |
if __name__ == "__main__":
|
91 |
-
demo.launch()
|
|
|
3 |
import torch
|
4 |
from peft import PeftModel
|
5 |
import os
|
6 |
+
import csv
|
7 |
+
import huggingface_hub
|
8 |
+
from huggingface_hub import Repository, hf_hub_download, upload_file
|
9 |
+
from datetime import datetime
|
10 |
|
11 |
+
DATASET_REPO_URL = "https://huggingface.co/datasets/JerniganLab/chat-data"
|
12 |
+
DATASET_REPO_ID = "JerniganLab/chat-data"
|
13 |
+
DATA_FILENAME = "data.csv"
|
14 |
+
DATA_FILE = os.path.join("data", DATA_FILENAME)
|
15 |
HF_TOKEN = os.environ.get("HF_TOKEN")
|
16 |
|
17 |
+
|
18 |
+
HF_TOKEN = os.environ.get("HF_TOKEN")
|
19 |
+
|
20 |
+
# overriding/appending to the gradio template
|
21 |
+
SCRIPT = """
|
22 |
+
<script>
|
23 |
+
if (!window.hasBeenRun) {
|
24 |
+
window.hasBeenRun = true;
|
25 |
+
console.log("should only happen once");
|
26 |
+
document.querySelector("button.submit").click();
|
27 |
+
}
|
28 |
+
</script>
|
29 |
+
"""
|
30 |
+
with open(os.path.join(gr.routes.STATIC_TEMPLATE_LIB, "frontend", "index.html"), "a") as f:
|
31 |
+
f.write(SCRIPT)
|
32 |
+
|
33 |
+
try:
|
34 |
+
hf_hub_download(
|
35 |
+
repo_id=DATASET_REPO_ID,
|
36 |
+
filename=DATA_FILENAME,
|
37 |
+
cache_dir=DATA_DIRNAME,
|
38 |
+
repo_type='dataset',
|
39 |
+
force_filename=DATA_FILENAME
|
40 |
+
)
|
41 |
+
except:
|
42 |
+
print("file not found")
|
43 |
+
|
44 |
+
repo = Repository(
|
45 |
+
local_dir="data", clone_from=DATASET_REPO_URL, use_auth_token=HF_TOKEN
|
46 |
+
)
|
47 |
+
|
48 |
+
|
49 |
+
|
50 |
+
|
51 |
model_id = "JerniganLab/interviews-and-qa"
|
52 |
base_model = "meta-llama/Meta-Llama-3-8B-Instruct"
|
53 |
|
|
|
64 |
|
65 |
pipeline.model = PeftModel.from_pretrained(llama_model, model_id)
|
66 |
|
67 |
+
def store_message(message: str, system_prompt: str, response: str):
|
68 |
+
if response and message:
|
69 |
+
with open(DATA_FILE, "a") as csvfile:
|
70 |
+
writer = csv.DictWriter(csvfile, fieldnames=["message","system_prompt","response","time"])
|
71 |
+
writer.writerow(
|
72 |
+
{"message": message, "system_prompt": system_prompt, "response": response, "time": str(datetime.now())}
|
73 |
+
)
|
74 |
+
commit_url = repo.push_to_hub()
|
75 |
+
# return generate_html()
|
76 |
+
|
77 |
+
|
78 |
# def chat_function(message, history, system_prompt, max_new_tokens, temperature):
|
79 |
# messages = [{"role":"system","content":system_prompt},
|
80 |
# {"role":"user", "content":message}]
|
|
|
112 |
do_sample = True,
|
113 |
temperature = temperature + 0.1,
|
114 |
top_p = 0.9,)
|
115 |
+
store_message(message, system_prompt, outputs[0]["generated_text"][len(prompt):])
|
116 |
return outputs[0]["generated_text"][len(prompt):]
|
117 |
|
118 |
"""
|
|
|
136 |
additional_inputs=[
|
137 |
gr.Slider(100,4000, label="Max New Tokens"),
|
138 |
gr.Slider(0,1, label="Temperature")
|
139 |
+
],
|
140 |
+
type="messages",
|
141 |
+
save_history=True,
|
142 |
)
|
143 |
|
144 |
|
145 |
|
146 |
if __name__ == "__main__":
|
147 |
+
demo.launch()
|