Update README.md
Browse files
README.md
CHANGED
@@ -12,3 +12,77 @@ This model is part of my thesis to make a question-answering bot for SUNY Brockp
|
|
12 |
|
13 |
For more information, see my [github repo](https://github.com/msaad02/honors-thesis) for the project.
|
14 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
|
13 |
For more information, see my [github repo](https://github.com/msaad02/honors-thesis) for the project.
|
14 |
|
15 |
+
## Examples
|
16 |
+
|
17 |
+
>Are SAT scores required?
|
18 |
+
>>SAT scores are not required for admission, but test scores are used as part of Brockport's holistic review process to make admissions decisions.
|
19 |
+
|
20 |
+
>Who can I contact for help with financial aid?
|
21 |
+
>>You can contact the financial aid office at SUNY Brockport for assistance with financial aid. They are available to answer any questions and help you navigate the financial aid process.
|
22 |
+
|
23 |
+
|
24 |
+
## Prompting
|
25 |
+
|
26 |
+
This prompting style is what was used for finetuning:
|
27 |
+
|
28 |
+
>Below is an inquiry related to SUNY Brockport - from academics, admissions, and faculty support to student life. Prioritize accuracy and brevity.
|
29 |
+
>
|
30 |
+
>\### Instruction:\
|
31 |
+
>{question}
|
32 |
+
>
|
33 |
+
>\### Response:\
|
34 |
+
>{response}
|
35 |
+
|
36 |
+
## Usage
|
37 |
+
|
38 |
+
To run this model, I suggest using the following code to build a pipeline, which implements the correct prompt formatting
|
39 |
+
|
40 |
+
```python
|
41 |
+
import torch
|
42 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
|
43 |
+
import textwrap
|
44 |
+
|
45 |
+
bnb_config = BitsAndBytesConfig(
|
46 |
+
load_in_4bit=True,
|
47 |
+
bnb_4bit_quant_type="nf4",
|
48 |
+
bnb_4bit_compute_dtype=torch.bfloat16,
|
49 |
+
)
|
50 |
+
|
51 |
+
model = AutoModelForCausalLM.from_pretrained(
|
52 |
+
"msaad02/llama2_7b_brockport_gpt",
|
53 |
+
quantization_config=bnb_config,
|
54 |
+
device_map={"": 0},
|
55 |
+
trust_remote_code=True
|
56 |
+
)
|
57 |
+
|
58 |
+
tokenizer = AutoTokenizer.from_pretrained("msaad02/llama2_7b_brockport_gpt")
|
59 |
+
|
60 |
+
# Use a pipeline as a high-level helper
|
61 |
+
from transformers import pipeline
|
62 |
+
|
63 |
+
pipe = pipeline(
|
64 |
+
task="text-generation",
|
65 |
+
model=model,
|
66 |
+
tokenizer=tokenizer
|
67 |
+
)
|
68 |
+
|
69 |
+
def qa(text: str, full = False):
|
70 |
+
# textwrap.dedent gets rid of indenting at the start of each newline
|
71 |
+
text = textwrap.dedent(f"""\
|
72 |
+
Below is an inquiry related to SUNY Brockport - from academics, admissions, and faculty support to student life. Prioritize accuracy and brevity.
|
73 |
+
|
74 |
+
### Instruction:
|
75 |
+
{text}
|
76 |
+
|
77 |
+
### Response:
|
78 |
+
""")
|
79 |
+
|
80 |
+
response = pipe(text, max_length=100, do_sample=True, top_k=50, top_p=0.95, temperature=1.0)
|
81 |
+
response = response[0]['generated_text']
|
82 |
+
response = response.split("### Response:\n")[1] if not full else response
|
83 |
+
|
84 |
+
return response
|
85 |
+
|
86 |
+
print(qa("How do I apply?", full = True))
|
87 |
+
```
|
88 |
+
|