andrewzamai
commited on
Commit
•
30be815
1
Parent(s):
0979f85
Update README.md
Browse files
README.md
CHANGED
@@ -13,6 +13,63 @@ SLIMER is an LLM specifically instructed for zero-shot NER.
|
|
13 |
|
14 |
Instructed on a reduced number of samples, it is designed to tackle never-seen-before Named Entity tags by leveraging a prompt enriched with a DEFINITION and GUIDELINES for the NE to be extracted.
|
15 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
16 |
<img src="https://huggingface.co/expertai/SLIMER/resolve/main/SLIMER_instruction_prompt.png" width="200">
|
17 |
|
18 |
Currently existing approaches fine-tune on an extensive number of entity classes (around 13K) and assess zero-shot NER capabilities on Out-Of-Distribution input domains.
|
|
|
13 |
|
14 |
Instructed on a reduced number of samples, it is designed to tackle never-seen-before Named Entity tags by leveraging a prompt enriched with a DEFINITION and GUIDELINES for the NE to be extracted.
|
15 |
|
16 |
+
<!DOCTYPE html>
|
17 |
+
<html>
|
18 |
+
<head>
|
19 |
+
<style>
|
20 |
+
.container {
|
21 |
+
border: 2px solid #000;
|
22 |
+
padding: 10px;
|
23 |
+
width: 600px;
|
24 |
+
margin: 0 auto;
|
25 |
+
font-family: Arial, sans-serif;
|
26 |
+
}
|
27 |
+
.header {
|
28 |
+
background-color: black;
|
29 |
+
color: white;
|
30 |
+
padding: 5px;
|
31 |
+
text-align: center;
|
32 |
+
font-weight: bold;
|
33 |
+
}
|
34 |
+
.content {
|
35 |
+
padding: 10px;
|
36 |
+
}
|
37 |
+
.definition {
|
38 |
+
background-color: #ccffcc;
|
39 |
+
padding: 5px;
|
40 |
+
}
|
41 |
+
.guidelines {
|
42 |
+
background-color: #ffffcc;
|
43 |
+
padding: 5px;
|
44 |
+
}
|
45 |
+
</style>
|
46 |
+
</head>
|
47 |
+
<body>
|
48 |
+
<div class="container">
|
49 |
+
<div class="header">Instruction Tuning Prompt</div>
|
50 |
+
<div class="content">
|
51 |
+
<p><b>[INST]</b></p>
|
52 |
+
<p>You are given a text chunk (delimited by triple quotes) and an instruction.<br>
|
53 |
+
Read the text and answer to the instruction in the end.</p>
|
54 |
+
<p>"""<br>
|
55 |
+
{input text}<br>
|
56 |
+
"""</p>
|
57 |
+
<p><b>Instruction:</b> Extract the Named Entities of type <b>DATE</b> from the text chunk you have read.</p>
|
58 |
+
<p>You are given a <b>DEFINITION</b> and some <b>GUIDELINES</b>.</p>
|
59 |
+
<div class="definition">
|
60 |
+
<p><b>DEFINITION:</b> <b>DATE</b> refers to specific points in time, including days, months, years, and relative time expressions like 'Week 2'.</p>
|
61 |
+
</div>
|
62 |
+
<div class="guidelines">
|
63 |
+
<p><b>GUIDELINES:</b> Avoid labeling non-specific time references like 'recently' or 'soon'. Exercise caution with ambiguous terms like 'May' (month or verb) and 'Wednesday Adams' (person's name which includes a day of the week).</p>
|
64 |
+
</div>
|
65 |
+
<p>Return a JSON list of instances of this Named Entity type. Return an empty list if no instances are present.</p>
|
66 |
+
<p><b>[/INST]</b></p>
|
67 |
+
</div>
|
68 |
+
</div>
|
69 |
+
</body>
|
70 |
+
</html>
|
71 |
+
|
72 |
+
|
73 |
<img src="https://huggingface.co/expertai/SLIMER/resolve/main/SLIMER_instruction_prompt.png" width="200">
|
74 |
|
75 |
Currently existing approaches fine-tune on an extensive number of entity classes (around 13K) and assess zero-shot NER capabilities on Out-Of-Distribution input domains.
|