rahulvk007
commited on
Commit
•
96a6287
1
Parent(s):
9a0aeee
Update README.md
Browse files
README.md
CHANGED
@@ -10,14 +10,66 @@ tags:
|
|
10 |
- llama
|
11 |
- trl
|
12 |
- sft
|
|
|
|
|
13 |
---
|
14 |
|
15 |
-
#
|
16 |
|
17 |
-
- **Developed by:** rahulvk007
|
18 |
-
- **License:**
|
19 |
-
- **
|
|
|
20 |
|
21 |
-
This
|
22 |
|
23 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
- llama
|
11 |
- trl
|
12 |
- sft
|
13 |
+
datasets:
|
14 |
+
- rahulvk007/quenumber_extraction_v2
|
15 |
---
|
16 |
|
17 |
+
# ExtractQueNumberMini Model
|
18 |
|
19 |
+
- **Developed by:** [rahulvk007](https://github.com/rahulvk007) ([rahulvk.com](https://www.rahulvk.com))
|
20 |
+
- **License:** [Apache-2.0](https://opensource.org/licenses/Apache-2.0)
|
21 |
+
- **Base Model:** [unsloth/SmolLM2-135M](https://huggingface.co/unsloth/SmolLM2-135M)
|
22 |
+
- **Finetuning**: Optimized with [Unsloth](https://github.com/unslothai/unsloth) and [Hugging Face's TRL library](https://github.com/huggingface/trl)
|
23 |
|
24 |
+
This model has been fine-tuned for quick extraction of question numbers from OCRed handwritten text. It is designed to run efficiently on CPU due to its compact size.
|
25 |
|
26 |
+
### Model Usage
|
27 |
+
|
28 |
+
To use this model, set the system prompt to the following:
|
29 |
+
|
30 |
+
> **Extract the question number from the given text. Your response should be just an integer representing the question number. Do not provide any explanation or context. Just the number.**
|
31 |
+
|
32 |
+
### Inference Code Example
|
33 |
+
|
34 |
+
```python
|
35 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
36 |
+
|
37 |
+
checkpoint = "rahulvk007/ExtractQueNumberMini"
|
38 |
+
device = "cpu" # change to "cuda" for GPU
|
39 |
+
|
40 |
+
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
|
41 |
+
model = AutoModelForCausalLM.from_pretrained(checkpoint).to(device)
|
42 |
+
|
43 |
+
alpaca_prompt = """Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.
|
44 |
+
|
45 |
+
### Instruction:
|
46 |
+
{}
|
47 |
+
|
48 |
+
### Input:
|
49 |
+
{}
|
50 |
+
|
51 |
+
### Response:
|
52 |
+
{}"""
|
53 |
+
|
54 |
+
inputs = tokenizer(
|
55 |
+
[
|
56 |
+
alpaca_prompt.format(
|
57 |
+
"Extract the question number from the given text. Your response should be just an integer which is the question number. Do not provide any explanation or context. Just the number.",
|
58 |
+
"<Give OCR Text here>",
|
59 |
+
"",
|
60 |
+
)
|
61 |
+
],
|
62 |
+
return_tensors="pt"
|
63 |
+
).to(device)
|
64 |
+
|
65 |
+
outputs = model.generate(**inputs, max_new_tokens=64, use_cache=True)
|
66 |
+
print(tokenizer.batch_decode(outputs, skip_special_tokens=True))
|
67 |
+
```
|
68 |
+
|
69 |
+
### Datasets
|
70 |
+
|
71 |
+
The model was fine-tuned on [rahulvk007/quenumber_extraction_v2](https://huggingface.co/datasets/rahulvk007/quenumber_extraction_v2), specifically curated for this task.
|
72 |
+
|
73 |
+
---
|
74 |
+
|
75 |
+
[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
|