Spaces:
Sleeping
Sleeping
fix: simplify model loading and generation
Browse files
app.py
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
import gradio as gr
|
2 |
-
from transformers import
|
3 |
import torch
|
4 |
import numpy as np
|
5 |
from monitoring import PerformanceMonitor, measure_time
|
@@ -13,33 +13,33 @@ monitor = PerformanceMonitor()
|
|
13 |
|
14 |
def format_prompt(problem):
|
15 |
"""Format the input problem according to the model's expected format"""
|
16 |
-
return f"
|
17 |
|
18 |
@measure_time
|
19 |
def get_model_response(problem, model_id):
|
20 |
"""Get response from a specific model"""
|
21 |
try:
|
22 |
-
# Initialize pipeline
|
23 |
pipe = pipeline(
|
24 |
"text-generation",
|
25 |
model=model_id,
|
26 |
torch_dtype=torch.float16,
|
27 |
device_map="auto",
|
|
|
28 |
)
|
29 |
|
30 |
# Format prompt and generate response
|
31 |
prompt = format_prompt(problem)
|
32 |
response = pipe(
|
33 |
prompt,
|
34 |
-
max_new_tokens=
|
35 |
temperature=0.1,
|
36 |
-
|
37 |
-
|
|
|
38 |
)[0]["generated_text"]
|
39 |
|
40 |
-
|
41 |
-
assistant_response = response.split("<|im_start|>assistant\n")[-1].split("<|im_end|>")[0]
|
42 |
-
return assistant_response.strip()
|
43 |
except Exception as e:
|
44 |
return f"Error: {str(e)}"
|
45 |
|
@@ -59,11 +59,24 @@ def solve_problem(problem, problem_type):
|
|
59 |
base_response, base_time = get_model_response(problem, BASE_MODEL_ID)
|
60 |
finetuned_response, finetuned_time = get_model_response(problem, FINETUNED_MODEL_ID)
|
61 |
|
62 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
63 |
monitor.record_response_time("base", base_time)
|
64 |
monitor.record_response_time("finetuned", finetuned_time)
|
65 |
-
|
66 |
-
# Record success (basic check - no error message)
|
67 |
monitor.record_success("base", not base_response.startswith("Error"))
|
68 |
monitor.record_success("finetuned", not finetuned_response.startswith("Error"))
|
69 |
|
@@ -82,12 +95,12 @@ def solve_problem(problem, problem_type):
|
|
82 |
- Base Model: {stats.get('base_success_rate', 0):.1f}%
|
83 |
- Fine-tuned Model: {stats.get('finetuned_success_rate', 0):.1f}%
|
84 |
|
85 |
-
#### Problem
|
86 |
"""
|
87 |
for ptype, percentage in stats.get('problem_type_distribution', {}).items():
|
88 |
stats_display += f"- {ptype}: {percentage:.1f}%\n"
|
89 |
|
90 |
-
return
|
91 |
|
92 |
# Create Gradio interface
|
93 |
with gr.Blocks(title="Mathematics Problem Solver") as demo:
|
@@ -98,12 +111,12 @@ with gr.Blocks(title="Mathematics Problem Solver") as demo:
|
|
98 |
with gr.Column():
|
99 |
problem_type = gr.Dropdown(
|
100 |
choices=["Addition", "Root Finding", "Derivative", "Custom"],
|
101 |
-
value="
|
102 |
label="Problem Type"
|
103 |
)
|
104 |
problem_input = gr.Textbox(
|
105 |
label="Enter your math problem",
|
106 |
-
placeholder="Example:
|
107 |
)
|
108 |
solve_btn = gr.Button("Solve", variant="primary")
|
109 |
|
@@ -123,9 +136,9 @@ with gr.Blocks(title="Mathematics Problem Solver") as demo:
|
|
123 |
# Example problems
|
124 |
gr.Examples(
|
125 |
examples=[
|
126 |
-
["
|
127 |
-
["
|
128 |
-
["
|
129 |
["\\sin{\\left(x\\right)}", "Derivative"],
|
130 |
["e^x", "Derivative"],
|
131 |
["\\frac{1}{x}", "Derivative"],
|
|
|
1 |
import gradio as gr
|
2 |
+
from transformers import pipeline
|
3 |
import torch
|
4 |
import numpy as np
|
5 |
from monitoring import PerformanceMonitor, measure_time
|
|
|
13 |
|
14 |
def format_prompt(problem):
|
15 |
"""Format the input problem according to the model's expected format"""
|
16 |
+
return f"Given a mathematical function, find its derivative.\n\nFunction: {problem}\nThe derivative of this function is:"
|
17 |
|
18 |
@measure_time
|
19 |
def get_model_response(problem, model_id):
|
20 |
"""Get response from a specific model"""
|
21 |
try:
|
22 |
+
# Initialize pipeline for each request
|
23 |
pipe = pipeline(
|
24 |
"text-generation",
|
25 |
model=model_id,
|
26 |
torch_dtype=torch.float16,
|
27 |
device_map="auto",
|
28 |
+
model_kwargs={"low_cpu_mem_usage": True}
|
29 |
)
|
30 |
|
31 |
# Format prompt and generate response
|
32 |
prompt = format_prompt(problem)
|
33 |
response = pipe(
|
34 |
prompt,
|
35 |
+
max_new_tokens=50, # Shorter response
|
36 |
temperature=0.1,
|
37 |
+
do_sample=False, # Deterministic
|
38 |
+
num_return_sequences=1,
|
39 |
+
return_full_text=False # Only return new text
|
40 |
)[0]["generated_text"]
|
41 |
|
42 |
+
return response.strip()
|
|
|
|
|
43 |
except Exception as e:
|
44 |
return f"Error: {str(e)}"
|
45 |
|
|
|
59 |
base_response, base_time = get_model_response(problem, BASE_MODEL_ID)
|
60 |
finetuned_response, finetuned_time = get_model_response(problem, FINETUNED_MODEL_ID)
|
61 |
|
62 |
+
# Format responses with steps
|
63 |
+
base_output = f"""Solution: {base_response}
|
64 |
+
|
65 |
+
Let's verify this step by step:
|
66 |
+
1. Starting with f(x) = {problem}
|
67 |
+
2. Applying differentiation rules
|
68 |
+
3. We get f'(x) = {base_response}"""
|
69 |
+
|
70 |
+
finetuned_output = f"""Solution: {finetuned_response}
|
71 |
+
|
72 |
+
Let's verify this step by step:
|
73 |
+
1. Starting with f(x) = {problem}
|
74 |
+
2. Applying differentiation rules
|
75 |
+
3. We get f'(x) = {finetuned_response}"""
|
76 |
+
|
77 |
+
# Record metrics
|
78 |
monitor.record_response_time("base", base_time)
|
79 |
monitor.record_response_time("finetuned", finetuned_time)
|
|
|
|
|
80 |
monitor.record_success("base", not base_response.startswith("Error"))
|
81 |
monitor.record_success("finetuned", not finetuned_response.startswith("Error"))
|
82 |
|
|
|
95 |
- Base Model: {stats.get('base_success_rate', 0):.1f}%
|
96 |
- Fine-tuned Model: {stats.get('finetuned_success_rate', 0):.1f}%
|
97 |
|
98 |
+
#### Problem Types Used
|
99 |
"""
|
100 |
for ptype, percentage in stats.get('problem_type_distribution', {}).items():
|
101 |
stats_display += f"- {ptype}: {percentage:.1f}%\n"
|
102 |
|
103 |
+
return base_output, finetuned_output, stats_display
|
104 |
|
105 |
# Create Gradio interface
|
106 |
with gr.Blocks(title="Mathematics Problem Solver") as demo:
|
|
|
111 |
with gr.Column():
|
112 |
problem_type = gr.Dropdown(
|
113 |
choices=["Addition", "Root Finding", "Derivative", "Custom"],
|
114 |
+
value="Derivative",
|
115 |
label="Problem Type"
|
116 |
)
|
117 |
problem_input = gr.Textbox(
|
118 |
label="Enter your math problem",
|
119 |
+
placeholder="Example: x^2 + 3x"
|
120 |
)
|
121 |
solve_btn = gr.Button("Solve", variant="primary")
|
122 |
|
|
|
136 |
# Example problems
|
137 |
gr.Examples(
|
138 |
examples=[
|
139 |
+
["x^2 + 3x", "Derivative"],
|
140 |
+
["144", "Root Finding"],
|
141 |
+
["235 + 567", "Addition"],
|
142 |
["\\sin{\\left(x\\right)}", "Derivative"],
|
143 |
["e^x", "Derivative"],
|
144 |
["\\frac{1}{x}", "Derivative"],
|