Update README.md
Browse files
README.md
CHANGED
@@ -62,51 +62,6 @@ for idx in range(1, len(example["completions"])+1):
|
|
62 |
# Step 9 Predicted (score): False (0.97) Label: False
|
63 |
```
|
64 |
|
65 |
-
Example 1)
|
66 |
-
|
67 |
-
```python
|
68 |
-
from datasets import load_dataset
|
69 |
-
from transformers import pipeline
|
70 |
-
import os
|
71 |
-
os.environ["HF_HUB_ENABLE_HF_TRANSFER"] = "1"
|
72 |
-
|
73 |
-
model_name = "plaguss/Qwen2.5-0.5B-Math-Shepherd-PRM-0.2"
|
74 |
-
|
75 |
-
pipe = pipeline("token-classification", model=model_name, device="cuda")
|
76 |
-
dataset = load_dataset("trl-lib/math_shepherd")
|
77 |
-
example = dataset["test"][10]
|
78 |
-
|
79 |
-
sep = "\n"
|
80 |
-
|
81 |
-
print(sep.join((example["prompt"], *example["completions"])))
|
82 |
-
for idx in range(1, len(example["completions"])+1):
|
83 |
-
text = sep.join((example["prompt"], *example["completions"][0:idx])) + sep
|
84 |
-
output = pipe(text)
|
85 |
-
score = float(output[-1]["score"])
|
86 |
-
pred = True if output[-1]["entity"] == "LABEL_1" else False
|
87 |
-
print(f"Step {idx}\tPredicted (score): {pred} ({score:.2f})\tLabel: {example['labels'][idx-1]}")
|
88 |
-
|
89 |
-
# Grandma gave Bryce and Carter some raisins. Bryce received 6 more raisins than Carter, and Carter received half the number of raisins Bryce received. How many raisins did Bryce receive?
|
90 |
-
# Step 1: Let $b$ be the number of raisins Bryce received and $c$ be the number of raisins Carter received.
|
91 |
-
# Step 2: We are given that $b = c + 6$ and $c = \frac{1}{2}b$.
|
92 |
-
# Step 3: Substituting the second equation into the first equation, we get $b = c + 6 = \frac{1}{2}b + 6$.
|
93 |
-
# Step 4: Simplifying, we have $b = \frac{1}{2}b + 6$.
|
94 |
-
# Step 5: Subtracting $\frac{1}{2}b$ from both sides, we get $\frac{1}{2}b - b = 6$.
|
95 |
-
# Step 6: Simplifying further, we have $\frac{1}{2}b - 2b = 6$.
|
96 |
-
# Step 7: Combining like terms, we have $-\frac{1}{2}b = 6$.
|
97 |
-
# Step 8: Multiplying both sides by $-2$, we get $b = -12$.
|
98 |
-
# Step 9: Therefore, Bryce received $\boxed{-12}$ raisins.The answer is: -12
|
99 |
-
# Step 1 Predicted (score): True (0.99) Label: True
|
100 |
-
# Step 2 Predicted (score): True (0.99) Label: True
|
101 |
-
# Step 3 Predicted (score): True (0.94) Label: True
|
102 |
-
# Step 4 Predicted (score): True (0.82) Label: True
|
103 |
-
# Step 5 Predicted (score): True (0.58) Label: True
|
104 |
-
# Step 6 Predicted (score): False (0.62) Label: False
|
105 |
-
# Step 7 Predicted (score): False (0.77) Label: False
|
106 |
-
# Step 8 Predicted (score): False (0.91) Label: False
|
107 |
-
# Step 9 Predicted (score): False (0.97) Label: False
|
108 |
-
```
|
109 |
-
|
110 |
Example 2)
|
111 |
|
112 |
```python
|
@@ -193,16 +148,16 @@ for i, example in enumerate(examples):
|
|
193 |
pred = True if output[-1]["entity"] == "LABEL_1" else False
|
194 |
print(f"Step {idx}\tPredicted (score): {pred} ({score:.2f})\tLabel: {example['labels'][idx-1]}")
|
195 |
|
196 |
-
- Example 0:
|
197 |
-
Step 1 Predicted (score): True (0.90) Label: True
|
198 |
-
Step 2 Predicted (score): False (0.55) Label: True
|
199 |
-
Step 3 Predicted (score): False (0.62) Label: True
|
200 |
-
Step 4 Predicted (score): False (0.90) Label: True
|
201 |
-
- Example 1:
|
202 |
-
Step 1 Predicted (score): True (0.90) Label: True
|
203 |
-
Step 2 Predicted (score): False (0.55) Label: True
|
204 |
-
Step 3 Predicted (score): False (0.62) Label: True
|
205 |
-
Step 4 Predicted (score): False (0.96) Label: False
|
206 |
```
|
207 |
|
208 |
|
|
|
62 |
# Step 9 Predicted (score): False (0.97) Label: False
|
63 |
```
|
64 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
65 |
Example 2)
|
66 |
|
67 |
```python
|
|
|
148 |
pred = True if output[-1]["entity"] == "LABEL_1" else False
|
149 |
print(f"Step {idx}\tPredicted (score): {pred} ({score:.2f})\tLabel: {example['labels'][idx-1]}")
|
150 |
|
151 |
+
# - Example 0:
|
152 |
+
# Step 1 Predicted (score): True (0.90) Label: True
|
153 |
+
# Step 2 Predicted (score): False (0.55) Label: True
|
154 |
+
# Step 3 Predicted (score): False (0.62) Label: True
|
155 |
+
# Step 4 Predicted (score): False (0.90) Label: True
|
156 |
+
# - Example 1:
|
157 |
+
# Step 1 Predicted (score): True (0.90) Label: True
|
158 |
+
# Step 2 Predicted (score): False (0.55) Label: True
|
159 |
+
# Step 3 Predicted (score): False (0.62) Label: True
|
160 |
+
# Step 4 Predicted (score): False (0.96) Label: False
|
161 |
```
|
162 |
|
163 |
|