Update app.py
Browse files
app.py
CHANGED
@@ -1,17 +1,29 @@
|
|
|
|
1 |
from transformers import AutoTokenizer, AutoModelForCausalLM
|
2 |
import torch
|
3 |
import gradio as gr
|
4 |
|
5 |
model_id = "witfoo/witq-1.0"
|
6 |
dtype = torch.float16 # float16 for Tesla T4, V100, bfloat16 for Ampere+
|
7 |
-
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
8 |
# Use GPU if available
|
9 |
-
device = torch.device("cuda" if torch.cuda.is_available() else "
|
10 |
-
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
15 |
|
16 |
|
17 |
preamble = "Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request."
|
|
|
1 |
+
## Chatbot for interacting with WitFoo's Opensource model with standard Transformers. Can run on GPU or CPU.
|
2 |
from transformers import AutoTokenizer, AutoModelForCausalLM
|
3 |
import torch
|
4 |
import gradio as gr
|
5 |
|
6 |
model_id = "witfoo/witq-1.0"
|
7 |
dtype = torch.float16 # float16 for Tesla T4, V100, bfloat16 for Ampere+
|
|
|
8 |
# Use GPU if available
|
9 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "auto")
|
10 |
+
|
11 |
+
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
12 |
+
try:
|
13 |
+
model = AutoModelForCausalLM.from_pretrained(
|
14 |
+
model_id,
|
15 |
+
torch_dtype=dtype,
|
16 |
+
device_map=device,
|
17 |
+
)
|
18 |
+
except:
|
19 |
+
if device == "cuda":
|
20 |
+
print("Failed to load model on GPU. Loading on CPU...")
|
21 |
+
device = "auto"
|
22 |
+
model = AutoModelForCausalLM.from_pretrained(
|
23 |
+
model_id,
|
24 |
+
torch_dtype=dtype,
|
25 |
+
device_map=device,
|
26 |
+
)
|
27 |
|
28 |
|
29 |
preamble = "Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request."
|