Update README.md
Browse files
README.md
CHANGED
@@ -6,7 +6,7 @@ language:
|
|
6 |
# Text Classification Toxicity
|
7 |
|
8 |
This is a quantized onnx model and is a fined-tuned version of [nreimers/MiniLMv2-L6-H384-distilled-from-BERT-Large](https://huggingface.co/nreimers/MiniLMv2-L6-H384-distilled-from-BERT-Large) on the on the [Jigsaw 1st Kaggle competition](https://www.kaggle.com/competitions/jigsaw-toxic-comment-classification-challenge) dataset using [unitary/toxic-bert](https://huggingface.co/unitary/toxic-bert) as teacher model.
|
9 |
-
The original model can be found [here](https://huggingface.co/minuva/MiniLMv2-toxic-
|
10 |
|
11 |
|
12 |
# Usage
|
@@ -16,7 +16,7 @@ The original model can be found [here](https://huggingface.co/minuva/MiniLMv2-to
|
|
16 |
```bash
|
17 |
pip install tokenizers
|
18 |
pip install onnxruntime
|
19 |
-
git clone https://huggingface.co/minuva/MiniLMv2-toxic-
|
20 |
```
|
21 |
|
22 |
|
@@ -31,7 +31,7 @@ from tokenizers import Tokenizer
|
|
31 |
from onnxruntime import InferenceSession
|
32 |
|
33 |
|
34 |
-
model_name = "minuva/MiniLMv2-toxic-
|
35 |
tokenizer = Tokenizer.from_pretrained(model_name)
|
36 |
tokenizer.enable_padding(
|
37 |
pad_token="<pad>",
|
@@ -42,9 +42,9 @@ batch_size = 16
|
|
42 |
|
43 |
texts = ["This is pure trash",]
|
44 |
outputs = []
|
45 |
-
model = InferenceSession("MiniLMv2-toxic-
|
46 |
|
47 |
-
with open(os.path.join("MiniLMv2-toxic-
|
48 |
config = json.load(f)
|
49 |
|
50 |
output_names = [output.name for output in model.get_outputs()]
|
@@ -115,7 +115,7 @@ The following hyperparameters were used during training:
|
|
115 |
|
116 |
| Teacher (params) | Student (params) | Set (metric) | Score (teacher) | Score (student) |
|
117 |
|--------------------|-------------|----------|--------| --------|
|
118 |
-
| unitary/toxic-bert (110M) | MiniLMv2-toxic-
|
119 |
|
120 |
# Deployment
|
121 |
|
|
|
6 |
# Text Classification Toxicity
|
7 |
|
8 |
This is a quantized onnx model and is a fined-tuned version of [nreimers/MiniLMv2-L6-H384-distilled-from-BERT-Large](https://huggingface.co/nreimers/MiniLMv2-L6-H384-distilled-from-BERT-Large) on the on the [Jigsaw 1st Kaggle competition](https://www.kaggle.com/competitions/jigsaw-toxic-comment-classification-challenge) dataset using [unitary/toxic-bert](https://huggingface.co/unitary/toxic-bert) as teacher model.
|
9 |
+
The original model can be found [here](https://huggingface.co/minuva/MiniLMv2-toxic-jigsaw)
|
10 |
|
11 |
|
12 |
# Usage
|
|
|
16 |
```bash
|
17 |
pip install tokenizers
|
18 |
pip install onnxruntime
|
19 |
+
git clone https://huggingface.co/minuva/MiniLMv2-toxic-jigsaw-onnx
|
20 |
```
|
21 |
|
22 |
|
|
|
31 |
from onnxruntime import InferenceSession
|
32 |
|
33 |
|
34 |
+
model_name = "minuva/MiniLMv2-toxic-jigsaw-onnx"
|
35 |
tokenizer = Tokenizer.from_pretrained(model_name)
|
36 |
tokenizer.enable_padding(
|
37 |
pad_token="<pad>",
|
|
|
42 |
|
43 |
texts = ["This is pure trash",]
|
44 |
outputs = []
|
45 |
+
model = InferenceSession("MiniLMv2-toxic-jigsaw-onnx/model_optimized_quantized.onnx", providers=['CUDAExecutionProvider'])
|
46 |
|
47 |
+
with open(os.path.join("MiniLMv2-toxic-jigsaw-onnx", "config.json"), "r") as f:
|
48 |
config = json.load(f)
|
49 |
|
50 |
output_names = [output.name for output in model.get_outputs()]
|
|
|
115 |
|
116 |
| Teacher (params) | Student (params) | Set (metric) | Score (teacher) | Score (student) |
|
117 |
|--------------------|-------------|----------|--------| --------|
|
118 |
+
| unitary/toxic-bert (110M) | MiniLMv2-toxic-jigsaw-onnx (23M) | Test (ROC_AUC) | 0.98636 | 0.98130 |
|
119 |
|
120 |
# Deployment
|
121 |
|