Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -1,35 +1,66 @@
|
|
1 |
import streamlit as st
|
2 |
-
from transformers import AutoTokenizer, AutoModelForCausalLM
|
3 |
-
from huggingface_hub import login
|
4 |
import torch
|
|
|
5 |
import os
|
|
|
|
|
6 |
|
|
|
|
|
|
|
|
|
7 |
|
8 |
-
#
|
9 |
model_map = {
|
10 |
-
"FinGPT": "AI4Finance/FinGPT",
|
11 |
-
"
|
12 |
-
"
|
|
|
|
|
13 |
}
|
14 |
|
15 |
-
# Cache
|
16 |
@st.cache_resource
|
17 |
-
def
|
18 |
-
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
19 |
model = AutoModelForCausalLM.from_pretrained(
|
20 |
model_id,
|
21 |
-
torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32
|
|
|
|
|
22 |
)
|
23 |
return model, tokenizer
|
24 |
|
25 |
-
#
|
26 |
-
def
|
27 |
-
model, tokenizer =
|
28 |
-
inputs = tokenizer(
|
29 |
outputs = model.generate(**inputs, max_new_tokens=150)
|
30 |
return tokenizer.decode(outputs[0], skip_special_tokens=True)
|
31 |
|
32 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
33 |
st.title("💼 Financial LLM Evaluation Interface")
|
34 |
|
35 |
model_choice = st.selectbox("Select a Financial Model", list(model_map.keys()))
|
@@ -38,7 +69,8 @@ user_question = st.text_area("Enter your financial question:", "What is EBITDA?"
|
|
38 |
if st.button("Get Response"):
|
39 |
with st.spinner("Generating response..."):
|
40 |
try:
|
41 |
-
|
|
|
42 |
st.subheader(f"Response from {model_choice}:")
|
43 |
st.write(answer)
|
44 |
except Exception as e:
|
|
|
1 |
import streamlit as st
|
|
|
|
|
2 |
import torch
|
3 |
+
import requests
|
4 |
import os
|
5 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM
|
6 |
+
from huggingface_hub import login, HfApi
|
7 |
|
8 |
+
# Optional: Login if you want access to gated/private models
|
9 |
+
HF_TOKEN = os.getenv("HF_TOKEN", None)
|
10 |
+
if HF_TOKEN:
|
11 |
+
login(HF_TOKEN)
|
12 |
|
13 |
+
# Define model map with access type
|
14 |
model_map = {
|
15 |
+
"FinGPT": {"id": "AI4Finance/FinGPT", "local": True},
|
16 |
+
"InvestLM": {"id": "mrm8488/investLM-7B", "local": False}, # example ID, update if needed
|
17 |
+
"FinLLaMA": {"id": "HuggingFaceH4/fin-llama", "local": False},
|
18 |
+
"FinanceConnect": {"id": "ceadar-ie/FinanceConnect-13B", "local": True},
|
19 |
+
"Sujet-Finance": {"id": "sujet-ai/Sujet-Finance-8B-v0.1", "local": True}
|
20 |
}
|
21 |
|
22 |
+
# Cache local models
|
23 |
@st.cache_resource
|
24 |
+
def load_local_model(model_id):
|
25 |
+
tokenizer = AutoTokenizer.from_pretrained(model_id, use_auth_token=HF_TOKEN)
|
26 |
model = AutoModelForCausalLM.from_pretrained(
|
27 |
model_id,
|
28 |
+
torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
|
29 |
+
device_map="auto" if torch.cuda.is_available() else None,
|
30 |
+
use_auth_token=HF_TOKEN
|
31 |
)
|
32 |
return model, tokenizer
|
33 |
|
34 |
+
# Local model querying
|
35 |
+
def query_local_model(model_id, prompt):
|
36 |
+
model, tokenizer = load_local_model(model_id)
|
37 |
+
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
|
38 |
outputs = model.generate(**inputs, max_new_tokens=150)
|
39 |
return tokenizer.decode(outputs[0], skip_special_tokens=True)
|
40 |
|
41 |
+
# Remote model querying (via Inference API)
|
42 |
+
def query_remote_model(model_id, prompt):
|
43 |
+
headers = {"Authorization": f"Bearer {HF_TOKEN}"} if HF_TOKEN else {}
|
44 |
+
payload = {"inputs": prompt, "parameters": {"max_new_tokens": 150}}
|
45 |
+
response = requests.post(
|
46 |
+
f"https://api-inference.huggingface.co/models/{model_id}",
|
47 |
+
headers=headers,
|
48 |
+
json=payload
|
49 |
+
)
|
50 |
+
if response.status_code == 200:
|
51 |
+
result = response.json()
|
52 |
+
return result[0]["generated_text"] if isinstance(result, list) else result.get("generated_text", "No output")
|
53 |
+
else:
|
54 |
+
raise RuntimeError(f"Failed to call remote model: {response.text}")
|
55 |
+
|
56 |
+
# Unified query dispatcher
|
57 |
+
def query_model(model_entry, prompt):
|
58 |
+
if model_entry["local"]:
|
59 |
+
return query_local_model(model_entry["id"], prompt)
|
60 |
+
else:
|
61 |
+
return query_remote_model(model_entry["id"], prompt)
|
62 |
+
|
63 |
+
# --- Streamlit UI ---
|
64 |
st.title("💼 Financial LLM Evaluation Interface")
|
65 |
|
66 |
model_choice = st.selectbox("Select a Financial Model", list(model_map.keys()))
|
|
|
69 |
if st.button("Get Response"):
|
70 |
with st.spinner("Generating response..."):
|
71 |
try:
|
72 |
+
model_entry = model_map[model_choice]
|
73 |
+
answer = query_model(model_entry, user_question)
|
74 |
st.subheader(f"Response from {model_choice}:")
|
75 |
st.write(answer)
|
76 |
except Exception as e:
|