Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -32,49 +32,45 @@ system_prompt = ("Adopt the persona of hilariously pissed off Andrej Karpathy "
|
|
32 |
"while always answering questions in full first principles analysis type of thinking "
|
33 |
"without using any analogies and always showing full working code or output in his answers.")
|
34 |
|
35 |
-
# Define a chat template
|
36 |
-
chat_template = {
|
37 |
-
"system": "<|system|>{content}<|end|>",
|
38 |
-
"user": "<|user|>{content}<|end|>",
|
39 |
-
"assistant": "<|assistant|>{content}<|end|>",
|
40 |
-
}
|
41 |
|
42 |
@spaces.GPU
|
43 |
def generate_response(message, history, temperature, max_new_tokens):
|
44 |
if model is None or tokenizer is None:
|
45 |
return "Model or tokenizer not loaded properly. Please check the logs."
|
46 |
|
47 |
-
|
48 |
-
|
49 |
|
50 |
-
inputs = tokenizer
|
51 |
|
52 |
with torch.no_grad():
|
53 |
generate_ids = model.generate(
|
54 |
-
inputs,
|
55 |
max_new_tokens=max_new_tokens,
|
56 |
do_sample=True,
|
57 |
temperature=temperature,
|
58 |
eos_token_id=tokenizer.eos_token_id,
|
59 |
)
|
60 |
-
response = tokenizer.decode(generate_ids[0, inputs.shape[1]:], skip_special_tokens=True)
|
61 |
return response.strip()
|
62 |
|
63 |
css = """
|
64 |
#output {
|
65 |
-
height:
|
66 |
overflow: auto;
|
67 |
-
border:
|
68 |
}
|
69 |
"""
|
70 |
|
71 |
with gr.Blocks(css=css) as demo:
|
72 |
gr.Markdown("# Nisten's Karpathy Chatbot with OSS OLMoE")
|
73 |
chatbot = gr.Chatbot(elem_id="output")
|
74 |
-
msg = gr.Textbox(label="
|
75 |
with gr.Row():
|
76 |
temperature = gr.Slider(minimum=0.1, maximum=1.0, value=0.7, step=0.1, label="Temperature")
|
77 |
-
max_new_tokens = gr.Slider(minimum=50, maximum=4000, value=
|
78 |
clear = gr.Button("Clear")
|
79 |
|
80 |
def user(user_message, history):
|
@@ -93,4 +89,4 @@ with gr.Blocks(css=css) as demo:
|
|
93 |
|
94 |
if __name__ == "__main__":
|
95 |
demo.queue(api_open=True)
|
96 |
-
demo.launch(debug=True, show_api=True, share=True)
|
|
|
32 |
"while always answering questions in full first principles analysis type of thinking "
|
33 |
"without using any analogies and always showing full working code or output in his answers.")
|
34 |
|
35 |
+
# Define a chat template as a string
|
36 |
+
chat_template = "<|system|>{system_message}<|end|><|user|>{user_message}<|end|><|assistant|>"
|
|
|
|
|
|
|
|
|
37 |
|
38 |
@spaces.GPU
|
39 |
def generate_response(message, history, temperature, max_new_tokens):
|
40 |
if model is None or tokenizer is None:
|
41 |
return "Model or tokenizer not loaded properly. Please check the logs."
|
42 |
|
43 |
+
# Construct the full prompt
|
44 |
+
full_prompt = chat_template.format(system_message=system_prompt, user_message=message)
|
45 |
|
46 |
+
inputs = tokenizer(full_prompt, return_tensors="pt").to(DEVICE)
|
47 |
|
48 |
with torch.no_grad():
|
49 |
generate_ids = model.generate(
|
50 |
+
inputs.input_ids,
|
51 |
max_new_tokens=max_new_tokens,
|
52 |
do_sample=True,
|
53 |
temperature=temperature,
|
54 |
eos_token_id=tokenizer.eos_token_id,
|
55 |
)
|
56 |
+
response = tokenizer.decode(generate_ids[0, inputs.input_ids.shape[1]:], skip_special_tokens=True)
|
57 |
return response.strip()
|
58 |
|
59 |
css = """
|
60 |
#output {
|
61 |
+
height: 1000px;
|
62 |
overflow: auto;
|
63 |
+
border: 2px solid #ccc;
|
64 |
}
|
65 |
"""
|
66 |
|
67 |
with gr.Blocks(css=css) as demo:
|
68 |
gr.Markdown("# Nisten's Karpathy Chatbot with OSS OLMoE")
|
69 |
chatbot = gr.Chatbot(elem_id="output")
|
70 |
+
msg = gr.Textbox(label="Meow")
|
71 |
with gr.Row():
|
72 |
temperature = gr.Slider(minimum=0.1, maximum=1.0, value=0.7, step=0.1, label="Temperature")
|
73 |
+
max_new_tokens = gr.Slider(minimum=50, maximum=4000, value=2000, step=50, label="Max New Tokens")
|
74 |
clear = gr.Button("Clear")
|
75 |
|
76 |
def user(user_message, history):
|
|
|
89 |
|
90 |
if __name__ == "__main__":
|
91 |
demo.queue(api_open=True)
|
92 |
+
demo.launch(debug=True, show_api=True, share=True)
|