Update README.md
Browse files
README.md
CHANGED
@@ -11,7 +11,7 @@ metrics:
|
|
11 |
|
12 |
<!-- Provide a quick summary of what the model is/does. -->
|
13 |
|
14 |
-
We use
|
15 |
|
16 |
### Model Description
|
17 |
|
@@ -47,7 +47,7 @@ When given a question, the prompt "System of linear equations: (Do not simplify)
|
|
47 |
|
48 |
```python
|
49 |
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
|
50 |
-
checkpoint = "zhuxunyu/
|
51 |
device = "cuda" # for GPU usage or "cpu" for CPU usage
|
52 |
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
|
53 |
model = AutoModelForSeq2SeqLM.from_pretrained(checkpoint).to(device)
|
@@ -62,7 +62,7 @@ generation = tokenizer.decode(output, skip_special_tokens=True)
|
|
62 |
|
63 |
```python
|
64 |
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
|
65 |
-
checkpoint = "zhuxunyu/
|
66 |
device = "cuda" # for GPU usage or "cpu" for CPU usage
|
67 |
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
|
68 |
model = AutoModelForSeq2SeqLM.from_pretrained(checkpoint).to(device)
|
@@ -76,7 +76,7 @@ generation = tokenizer.decode(output, skip_special_tokens=True)
|
|
76 |
|
77 |
```python
|
78 |
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
|
79 |
-
checkpoint = "zhuxunyu/
|
80 |
device = "cuda" # for GPU usage or "cpu" for CPU usage
|
81 |
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
|
82 |
model = AutoModelForSeq2SeqLM.from_pretrained(checkpoint).to(device)
|
@@ -111,7 +111,7 @@ their corresponding reasoning processes are built as a training dataset, and we
|
|
111 |
| PoT | 50.34 | 55.2 | 51.6 | 88.33 |
|
112 |
| EoT | 48.21 | 52.81 | 55.7 | 70.16 |
|
113 |
| CoT | 25.47 | 29.67 | 23.3 | 46.5 |
|
114 |
-
|
|
115 |
|
116 |
|
117 |
|
|
|
11 |
|
12 |
<!-- Provide a quick summary of what the model is/does. -->
|
13 |
|
14 |
+
We use Ensemble Thoughts Distillation to distill mathematical reasoning ability from gpt-3.5-turbo to CodeT5+-770m-py.
|
15 |
|
16 |
### Model Description
|
17 |
|
|
|
47 |
|
48 |
```python
|
49 |
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
|
50 |
+
checkpoint = "zhuxunyu/etd-codet5p-770m-py"
|
51 |
device = "cuda" # for GPU usage or "cpu" for CPU usage
|
52 |
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
|
53 |
model = AutoModelForSeq2SeqLM.from_pretrained(checkpoint).to(device)
|
|
|
62 |
|
63 |
```python
|
64 |
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
|
65 |
+
checkpoint = "zhuxunyu/etd-codet5p-770m-py"
|
66 |
device = "cuda" # for GPU usage or "cpu" for CPU usage
|
67 |
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
|
68 |
model = AutoModelForSeq2SeqLM.from_pretrained(checkpoint).to(device)
|
|
|
76 |
|
77 |
```python
|
78 |
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
|
79 |
+
checkpoint = "zhuxunyu/etd-codet5p-770m-py"
|
80 |
device = "cuda" # for GPU usage or "cpu" for CPU usage
|
81 |
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
|
82 |
model = AutoModelForSeq2SeqLM.from_pretrained(checkpoint).to(device)
|
|
|
111 |
| PoT | 50.34 | 55.2 | 51.6 | 88.33 |
|
112 |
| EoT | 48.21 | 52.81 | 55.7 | 70.16 |
|
113 |
| CoT | 25.47 | 29.67 | 23.3 | 46.5 |
|
114 |
+
| Ensemble_all | 50.56 | 55.34 | 52.3 | 88.83 |
|
115 |
|
116 |
|
117 |
|