Court First, Cash Later: WazirX Eyes Funds Dispersal in May End

TheCryptoTimesDipublikasikan tanggal 2025-04-22Terakhir diperbarui pada 2025-04-22

Nine months after the Rs 2000 crore cyber hack of WazirX cryptocurrency exchange, the logjam situation appears to have finally inched towards a closure, as the exchange users have overwhelmingly voted in favor of the management’s restructuring scheme.

Recently, the exchange management took to their official handle to post that the first distribution of user funds will initiate in the April-May 2025 window. However, it is already April 22, at the time of publishing of this report and going by the current circumstances, users are looking at May end for any chances of getting their funds back.

The WazirX management will be approaching the Singapore High Court on May 13 to get their restructuring scheme passed, after which the first dispersal process will only initiate under 10 business days.

The green light by Singapore Court will allow WazirX crypto exchange to finally start the funds return process.

What’s cooking at WazirX?

After the WazirX cryptocurrency exchange was hacked for Rs 2000 crore on July 18, 2024, the exchange management had suspended trading and kept on hold the remaining 55% of user funds, worth around Rs 2500 crore. Five months later, the exchange approached the Singapore High Court and drafted a restructuring proposal.

Acting on the directives of the court, the exchange management conducted a week long voting process of its users, in which they claimed to have received 93.4% votes in majority of their restructuring scheme.

Now, as per WazirX management, after the court approves their scheme, all users will be eligible to get their portfolio restored by 85%, at the time of the hack.

So, going by WazirX’s own calculation, funds distribution, if at all plausible, will begin only in May end.

So, if you’re eagerly awaiting those withdrawals, keep your fingers crossed for May 13—and maybe stock up on patience, because this “April-May” window may just take you all the way to June.

Also Read: Rise and Fall of WazirX: Mapping India’s Biggest Crypto Hack



Bacaan Terkait

AI Pembuatan Gambar Tanpa Pelatihan Dipercepat 1000%, Caranya: 'Pipa Tiga Tahap' Paling Sederhana

Kemampuan gambar AI semakin kuat, namun pengguna masih merasakannya lambat. Metode akselerasi model difusi tradisional seperti kuantisasi atau distilasi langkah sering kali bergantung pada perangkat keras atau fine-tuning yang mahal. Tim peneliti dari Beihang University, NTU, dan ETH memperkenalkan **MrFlow (Multi-Resolution Flow Matching)**, sebuah pipeline tiga tahap sederhana dan bebas pelatihan untuk mempercepat pembuatan gambar secara signifikan: 1. **Pembuatan Kerangka Beresolusi Rendah:** Model asli menghasilkan gambar struktur global (subjek, tata letak, semantik) di ruang latens beresolusi rendah. Token gambar jauh lebih sedikit, sehingga setiap langkah lebih murah dan konvergensinya lebih cepat. 2. **Super-Resolution di Ruang Pixel:** Hasil beresolusi rendah didekode ke gambar, lalu ditingkatkan resolusinya di ruang pixel menggunakan model super-resolution yang telah dilatih sebelumnya (seperti Real-ESRGAN). Pendekatan ini mempertahankan struktur dengan lebih baik daripada upsampling di ruang latens. 3. **Pemurnian Satu Langkah Beresolusi Tinggi:** Gambar super-resolution dienkode ulang ke ruang latens, ditambahkan sedikit noise intensitas rendah (~0.12), lalu dimurnikan oleh model flow-matching asli hanya dalam **satu langkah** inferensi resolusi tinggi. Noise rendah memungkinkan titik awal dekat dengan gambar bersih. Dengan konfigurasi default "12+1" (12 langkah rendah-res, 1 langkah tinggi-res), MrFlow mencapai **percepatan 10.35x** (dari 49.32s menjadi 4.77s) pada model seperti Qwen-Image, dengan penurunan kualitas minimal (~1%). Metode ini unggul dalam kurva trade-off kecepatan-kualitas dibanding metode akselerasi bebas pelatihan lainnya, dapat digabungkan dengan model distilasi untuk akselerasi lebih lanjut, dan sudah tersedia sebagai kode open-source beserta plugin ComfyUI.

marsbit1j yang lalu

AI Pembuatan Gambar Tanpa Pelatihan Dipercepat 1000%, Caranya: 'Pipa Tiga Tahap' Paling Sederhana

marsbit1j yang lalu

Trading

Spot
活动图片