Skip to content

Commit 10acc19

Browse files
ClaudioSavelliClaudioSavelli
committed
dataset release
1 parent 017c64f commit 10acc19

3 files changed

Lines changed: 70 additions & 23 deletions

File tree

‎awards.html‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -30,7 +30,7 @@
3030
<div class="banner" style="background: url('img/bari.jpg') no-repeat center; background-size: cover; height: 500px;">
3131
<div class="banner-table flex-column" style="background-color: rgba(0, 0, 0, 0.75);">
3232
<div class="flex-row">
33-
<div class="flex-item flex-column">
33+
<div class="flex-item flex-column">s
3434
<h1 class="add-top-margin-small strokeme">
3535
Task - Paper Awards
3636
</h1>

‎index.html‎

Lines changed: 32 additions & 22 deletions
Original file line numberDiff line numberDiff line change
@@ -78,28 +78,38 @@ <h3>Important Dates</h3>
7878

7979

8080
<h3>News</h3>
81-
<hr>
82-
<ul>
83-
<li>
84-
<p class="text">
85-
<b>Registration is open!</b></br>
86-
You can now register for the challenge by filling out the form
87-
<a href="https://docs.google.com/forms/d/1Vm3GD6NoyGbwq6vSS6oEFKTFbdp2j-thQSUox0sptXo/edit" target="_blank">here</a>.
88-
</p>
89-
</li>
90-
<li>
91-
<p class="text">
92-
<b>SVELA task announced!</b></br>
93-
The SVELA task has been officially announced as part of EVALITA 2026 Evaluation Campaign.
94-
</p>
95-
</li>
96-
<li>
97-
<p class="text">
98-
<b>Official task website is live!</b></br>
99-
The official website for SVELA @ EVALITA 2026 is now live.
100-
</p>
101-
</li>
102-
</ul>
81+
<hr>
82+
<ul>
83+
<li>
84+
<p class="text">
85+
<b>New dataset splits released!</b></br>
86+
The SVELA <a href="https://huggingface.co/datasets/ClaudioSavelli/SVELA-train-split" target="_blank">train split</a>
87+
(with retain, forget, and test partitions) and the
88+
<a href="https://huggingface.co/datasets/ClaudioSavelli/SVELA-val-split" target="_blank">validation split</a>
89+
(unlabeled) are now available on Hugging Face.
90+
</p>
91+
</li>
92+
<li>
93+
<p class="text">
94+
<b>Registration is open!</b></br>
95+
You can now register for the challenge by filling out the form
96+
<a href="https://docs.google.com/forms/d/1Vm3GD6NoyGbwq6vSS6oEFKTFbdp2j-thQSUox0sptXo/edit" target="_blank">here</a>.
97+
</p>
98+
</li>
99+
<li>
100+
<p class="text">
101+
<b>SVELA task announced!</b></br>
102+
The SVELA task has been officially announced as part of EVALITA 2026 Evaluation Campaign.
103+
</p>
104+
</li>
105+
<li>
106+
<p class="text">
107+
<b>Official task website is live!</b></br>
108+
The official website for SVELA @ EVALITA 2026 is now live.
109+
</p>
110+
</li>
111+
</ul>
112+
103113

104114

105115
<!-- <h3>Co-located with ICASSP 2025</h3>

‎workshop.html‎

Lines changed: 37 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -103,6 +103,43 @@ <h3>Multilingual Synthetic Dataset</h3>
103103
<p class="text">
104104
SVELA introduces the first multilingual synthetic dataset of fictional identities, ensuring that any knowledge the model shows comes solely from controlled fine-tuning, making forgetting measurable and reliable.
105105
</p>
106+
107+
108+
<!-- NEW SECTION: Dataset Splits Release -->
109+
<h3>Dataset Splits Release</h3>
110+
<p class="text">
111+
We have released official dataset splits on Hugging Face:
112+
</p>
113+
<ul>
114+
<li>
115+
<strong>SVELA – Train Split:</strong>
116+
<a href="https://huggingface.co/datasets/ClaudioSavelli/SVELA-train-split" target="_blank">view here</a>.<br>
117+
Contains identities, topic IDs, and questions with labels for <em>retain</em>, <em>forget</em>, and <em>test</em>.
118+
</li>
119+
<li>
120+
<strong>SVELA – Validation Split:</strong>
121+
<a href="https://huggingface.co/datasets/ClaudioSavelli/SVELA-val-split" target="_blank">view here</a>.<br>
122+
Larger set with the same structure but <em>unlabeled</em>, provided for validation.
123+
</li>
124+
</ul>
125+
126+
<p class="text">
127+
You can easily load the data using the 🤗 <code>datasets</code> library:
128+
</p>
129+
<pre><code class="language-python">from datasets import load_dataset
130+
131+
# --- Train split with labeled partitions ---
132+
retain = load_dataset("ClaudioSavelli/SVELA-train-split", split="retain")
133+
forget = load_dataset("ClaudioSavelli/SVELA-train-split", split="forget")
134+
test = load_dataset("ClaudioSavelli/SVELA-train-split", split="test")
135+
136+
# --- Unlabeled validation split ---
137+
val = load_dataset("ClaudioSavelli/SVELA-val-split", split="train")
138+
139+
print(retain.column_names)
140+
# ['identity_id', 'name', 'language', 'topic_id', 'question']</code></pre>
141+
<!-- END NEW SECTION -->
142+
106143

107144
<h3>Dataset Construction</h3>
108145
<p class="text">

0 commit comments

Comments
 (0)