Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
@@ -21,6 +21,10 @@ model = AutoModelForCausalLM.from_pretrained(
|
|
21 |
"meta-llama/Meta-Llama-3-8B-Instruct", quantization_config=quantization_config, token=token
|
22 |
)
|
23 |
tok = AutoTokenizer.from_pretrained("meta-llama/Meta-Llama-3-8B-Instruct", token=token)
|
|
|
|
|
|
|
|
|
24 |
|
25 |
if torch.cuda.is_available():
|
26 |
device = torch.device("cuda")
|
@@ -34,7 +38,7 @@ else:
|
|
34 |
|
35 |
|
36 |
@spaces.GPU(duration=150)
|
37 |
-
def chat(message, history, temperature,do_sample,
|
38 |
start_time = time.time()
|
39 |
chat = []
|
40 |
for item in history:
|
@@ -52,9 +56,13 @@ def chat(message, history, temperature,do_sample, top_k, max_tokens):
|
|
52 |
streamer=streamer,
|
53 |
max_new_tokens=max_tokens,
|
54 |
do_sample=True,
|
55 |
-
top_k=top_k,
|
56 |
temperature=temperature,
|
|
|
57 |
)
|
|
|
|
|
|
|
|
|
58 |
t = Thread(target=model.generate, kwargs=generate_kwargs)
|
59 |
t.start()
|
60 |
|
@@ -86,14 +94,11 @@ demo = gr.ChatInterface(
|
|
86 |
minimum=0, maximum=1, step=0.1, value=0.9, label="Temperature", render=False
|
87 |
),
|
88 |
gr.Checkbox(label="Sampling",value=True),
|
89 |
-
gr.Slider(
|
90 |
-
minimum=1, maximum=10000, step=5, value=1000, label="top_k", render=False
|
91 |
-
),
|
92 |
gr.Slider(
|
93 |
minimum=128,
|
94 |
maximum=4096,
|
95 |
step=1,
|
96 |
-
value=
|
97 |
label="Max new tokens",
|
98 |
render=False,
|
99 |
),
|
|
|
21 |
"meta-llama/Meta-Llama-3-8B-Instruct", quantization_config=quantization_config, token=token
|
22 |
)
|
23 |
tok = AutoTokenizer.from_pretrained("meta-llama/Meta-Llama-3-8B-Instruct", token=token)
|
24 |
+
terminators = [
|
25 |
+
tokenizer.eos_token_id,
|
26 |
+
tokenizer.convert_tokens_to_ids("<|eot_id|>")
|
27 |
+
]
|
28 |
|
29 |
if torch.cuda.is_available():
|
30 |
device = torch.device("cuda")
|
|
|
38 |
|
39 |
|
40 |
@spaces.GPU(duration=150)
|
41 |
+
def chat(message, history, temperature,do_sample, max_tokens):
|
42 |
start_time = time.time()
|
43 |
chat = []
|
44 |
for item in history:
|
|
|
56 |
streamer=streamer,
|
57 |
max_new_tokens=max_tokens,
|
58 |
do_sample=True,
|
|
|
59 |
temperature=temperature,
|
60 |
+
eos_token_id=terminators,
|
61 |
)
|
62 |
+
|
63 |
+
if temperature == 0:
|
64 |
+
generate_kwargs['do_sample'] = False
|
65 |
+
|
66 |
t = Thread(target=model.generate, kwargs=generate_kwargs)
|
67 |
t.start()
|
68 |
|
|
|
94 |
minimum=0, maximum=1, step=0.1, value=0.9, label="Temperature", render=False
|
95 |
),
|
96 |
gr.Checkbox(label="Sampling",value=True),
|
|
|
|
|
|
|
97 |
gr.Slider(
|
98 |
minimum=128,
|
99 |
maximum=4096,
|
100 |
step=1,
|
101 |
+
value=512,
|
102 |
label="Max new tokens",
|
103 |
render=False,
|
104 |
),
|