Update README.md
Browse files
README.md
CHANGED
@@ -38,19 +38,12 @@ E5-EG-small (E5 EverGreen - Small) is an efficient multilingual text classificat
|
|
38 |
## How to Get Started with the Model
|
39 |
|
40 |
```python
|
41 |
-
from transformers import
|
42 |
import torch
|
43 |
-
import time
|
44 |
|
45 |
# Load model and tokenizer
|
46 |
-
model_name = "s-nlp/E5-
|
47 |
-
|
48 |
-
model = AutoModelForSequenceClassification.from_pretrained(model_name)
|
49 |
-
|
50 |
-
# For optimal performance, use GPU if available
|
51 |
-
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
52 |
-
model = model.to(device)
|
53 |
-
model.eval()
|
54 |
|
55 |
# Batch classification example
|
56 |
questions = [
|
@@ -58,35 +51,13 @@ questions = [
|
|
58 |
"Who won the latest World Cup?",
|
59 |
"What is the speed of light?",
|
60 |
"What is the current Bitcoin price?"
|
|
|
|
|
61 |
]
|
62 |
|
63 |
-
# Tokenize all questions
|
64 |
-
inputs = tokenizer(
|
65 |
-
questions,
|
66 |
-
return_tensors="pt",
|
67 |
-
padding=True,
|
68 |
-
truncation=True,
|
69 |
-
max_length=64
|
70 |
-
).to(device)
|
71 |
-
|
72 |
# Classify
|
73 |
-
|
74 |
-
|
75 |
-
outputs = model(**inputs)
|
76 |
-
predictions = torch.nn.functional.softmax(outputs.logits, dim=-1)
|
77 |
-
predicted_classes = torch.argmax(predictions, dim=-1)
|
78 |
-
|
79 |
-
inference_time = (time.time() - start_time) * 1000 # ms
|
80 |
-
|
81 |
-
# Display results
|
82 |
-
class_names = ["Immutable", "Mutable"]
|
83 |
-
for i, question in enumerate(questions):
|
84 |
-
print(f"Q: {question}")
|
85 |
-
print(f" Classification: {class_names[predicted_classes[i].item()]}")
|
86 |
-
print(f" Confidence: {predictions[i][predicted_classes[i]].item():.2f}")
|
87 |
-
|
88 |
-
print(f"\nTotal inference time: {inference_time:.2f}ms")
|
89 |
-
print(f"Average per question: {inference_time/len(questions):.2f}ms")
|
90 |
```
|
91 |
|
92 |
## Training Details
|
|
|
38 |
## How to Get Started with the Model
|
39 |
|
40 |
```python
|
41 |
+
from transformers import pipeline
|
42 |
import torch
|
|
|
43 |
|
44 |
# Load model and tokenizer
|
45 |
+
model_name = "s-nlp/E5-EverGreen-Multilingual-Small"
|
46 |
+
pipe = pipeline("text-classification", model_name)
|
|
|
|
|
|
|
|
|
|
|
|
|
47 |
|
48 |
# Batch classification example
|
49 |
questions = [
|
|
|
51 |
"Who won the latest World Cup?",
|
52 |
"What is the speed of light?",
|
53 |
"What is the current Bitcoin price?"
|
54 |
+
"How old is Elon Musk",
|
55 |
+
"How old was Leo Tolstoy when he died?"
|
56 |
]
|
57 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
58 |
# Classify
|
59 |
+
results = pipe(questions)
|
60 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
61 |
```
|
62 |
|
63 |
## Training Details
|