ML Research Hub
32.9K subscribers
4.45K photos
273 videos
23 files
4.81K links
Advancing research in Machine Learning – practical insights, tools, and techniques for researchers.

Admin: @HusseinSheikho || @Hussein_Sheikho
Download Telegram
Forwarded from πŸ³
This media is not supported in your browser
VIEW IN TELEGRAM
🐳 @whale – #1 licensed platform gaming and sportsbook on Telegram!

1mil+ people trust us, 226k native users on @whalesocials, and the community is only growing!😈

❀️‍πŸ”₯Our buns ❀️‍πŸ”₯

πŸ₯°Supports BTC, USDT, TON, CELO and NOT
πŸ€‘Instant withdrawals
πŸ₯°Regular giveaways

Share your thoughts and feedback of @Whale on Ton.app and Trustpilot. You make us better ❔
Please open Telegram to view this post
VIEW IN TELEGRAM
πŸ‘3❀2❀‍πŸ”₯1
Flash Diffusion - acceleration of any diffusion model for image generation in a few steps

This is the name of the work recently published by ClΓ©ment Chadebec, Onur Tasar and their colleagues.
This is a distillation method for creating fast diffusion patterns.
When training the model using this method on the COCO2014 and COCO2017 datasets, the FID and CLIP-Score achieve good values; it requires only a few hours of training on the GPU and fewer model parameters than existing methods require.

🟑 Flash Diffusion page
πŸ–₯ GitHub

πŸ€— Flash SD
πŸ€— Flash SDXL

https://t.me/DataScienceT βœ…
Please open Telegram to view this post
VIEW IN TELEGRAM
Please open Telegram to view this post
VIEW IN TELEGRAM
πŸ‘2
πŸš€ AgentGym: Evolving Large Language Model-based Agents across Diverse Environments

πŸ–₯ Github: https://github.com/woooodyy/agentgym

πŸ”– Paper: https://arxiv.org/abs/2406.04151v1

β˜„οΈ Project: https://agentgym.github.io/

βœ”οΈ Model (AgentEvol-7B): https://huggingface.co/AgentGym/AgentEvol-7B

https://t.me/DataScienceT βœ…
Please open Telegram to view this post
VIEW IN TELEGRAM
Please open Telegram to view this post
VIEW IN TELEGRAM
πŸ‘2
This media is not supported in your browser
VIEW IN TELEGRAM
🎁 Lisa has given away over $100,000 in the last 30 days. Every single one of her subscribers is making money.

She is a professional trader and broadcasts her way of making money trading on her channel EVERY subscriber she has helped, and she will help you.

🧠 Do this and she will help you earn :

1. Subscribe to her channel
2. Write β€œGIFT” to her private messages
3. Follow her channel and trade with her. Repeat transactions after her = earn a lot of money.

Subscribe πŸ‘‡πŸ»
https://t.me/+VwDs29eEzRA5MTAx
πŸ‘2
⚑ L-MAGIC: Language Model Assisted Generation of Images with Coherence

The new Intel- L-MAGIC model can generate high-quality panoramic scenes based on a single input image and text prompt!

Numerous tests show that the model generates panoramic scenes with better scene placement and rendering quality than similar models.

β–ͺ️Github: https://github.com/IntelLabs/MMPano
β–ͺ️Paper: https://arxiv.org/abs/2406.01843
β–ͺ️Project: https://zhipengcai.github.io/MMPano/
β–ͺ️Video: https://youtu.be/XDMNEzH4-Ec

https://t.me/DataScienceT βœ…
Please open Telegram to view this post
VIEW IN TELEGRAM
Please open Telegram to view this post
VIEW IN TELEGRAM
πŸ‘7
πŸŽ™ VALLEY 2: Neural Codec Language Models are Human Parity Zero-Shot Text to Speech Synthesizers

In this article, Microsoft introduced VALL-E 2, the latest advancement in language models , which marks a major milestone in text-to-speech (TTS) synthesis, reaching the human level for the first time.

Experiments with LibriSpeech and VCTK datasets have shown that VALL-E 2 outperforms all previous models in terms of generated speech quality and naturalness.

βœ… Details: https://arxiv.org/abs/2406.05370
πŸ”΄ Demo of VALLE 2 will be available here: https://www.bing.com/?ref=aka&shorturl=valle2

https://t.me/DataScienceT βœ…
Please open Telegram to view this post
VIEW IN TELEGRAM
πŸ‘4❀3
This media is not supported in your browser
VIEW IN TELEGRAM
⭐️ New Dream Machine video generator from Luma AI.

Unlike Sora or KLING, it is available for testing.


You can try it here: https://lumalabs.ai/dream-machine

https://t.me/DataScienceT βœ…
Please open Telegram to view this post
VIEW IN TELEGRAM
πŸ‘4❀1
This media is not supported in your browser
VIEW IN TELEGRAM
🌟 MusicGPT is an application for running music-generating models locally

- brew install gabotechs/taps/musicgpt

MusicGPT allows you to run the latest music generation models locally on any platform, without installing heavy dependencies such as ML frameworks.

Currently, MusicGPT only supports MusicGen from Meta, but there are plans to add even more different music generation models.

Quick start with Docker:
docker run -it --gpus all -p 8642:8642 -v ~/.musicgpt:/root/.local/share/musicgpt gabotechs/musicgpt --gpu --ui-expose

or using cargo:
cargo install musicgpt

πŸ–₯ GitHub

https://t.me/DataScienceT βœ…
Please open Telegram to view this post
VIEW IN TELEGRAM
πŸ‘1
EU automatic quantitative system automatically searches for the lowest selling price of digital currencies such as BTC, ETH, USDT on the exchange, and buys in seconds.

1. Register and get 9999 USDT, with a maximum rebate of 10% deposit.

2. USDT quantification, funds are automatically credited and withdrawn.

3. Quantification VIP1-VIP8, quantitative income 3%-13%

4. Support multiple currencies, regular quantification 1.5%-6.0%

5. Quantification is reset once every 24 hours, and quantification is once a day.

6. Promote three-level quantitative trading agents (A5%+B2%+C1%=8% reward)

24-hour uninterrupted data collection, no need for manual observation, efficient and stable profit. This is the EU automatic quantitative system.

Telegram customer service: https://t.me/eu_online_service1

Member registration link: https://euexchange.cc/#/register?i=135623

EU official channel: https://t.me/EUExchangevip

EU official group: https://t.me/EUExchangeVipGroup
πŸ‘1
Forwarded from πŸ³
How to create passive income on Telegram?

You can make it with @Whale! πŸ₯°

The best part is that you can invite as many friends as you want and make tons of money while they play 🎲

What does your income consist of and how does it work?

🌟 You receive 10% of Whale's earnings from each direct referral.
🌟 1% for each 2nd level referral.
🌟 Monthly paid earnings in $TON.

The more friends you invite, the more chances you have to hit the big jackpot β€” get a share of the @whale jackpot when someone wins it! Sometimes it happens πŸ‘

Referrals are counted when:

βœ… Your friends follow your referral link.
βœ… Their wallets and Telegram accounts were not previously members of the Whale system.
βœ… They link their Telegram account to the bot.
βœ… They participate in some Whale games.

How to invite friends?

Get a unique invitation link by clicking β€œEarn” in the application itself or in the bot, and share this link with your friends! 🐳
Please open Telegram to view this post
VIEW IN TELEGRAM
πŸ‘3❀2❀‍πŸ”₯1
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ•Ή  VideoLLaMA 2 is a set of open-source Video-LLMs for video generation.

VideoLLaMA 2, a logical evolution of past models, includes a specialized space-time convolution (STC) component that effectively captures complex dynamics in video.

πŸ–₯ GitHub

πŸ€— Demo

βœ… VideoLLaMA 2 model

https://t.me/DataScienceT βœ…
Please open Telegram to view this post
VIEW IN TELEGRAM
πŸ‘9❀2❀‍πŸ”₯1πŸ‘Ž1πŸ†1
Forwarded from Advert Deal
πŸš€ 90% of people fail in #crypto because they pick the WRONG altcoins!

Be among those who know what to do! Harry spends countless hours researching altcoins that you SHOULD buy right now during this market dipπŸ”₯

πŸ’ΈDiscover altcoins with high growth potential! He shares all the information on his channel for free ⬇️ Subscribe now:
https://t.me/+-ewyDmzZceEwYzk0
https://t.me/+-ewyDmzZceEwYzk0
❀‍πŸ”₯4πŸ‘3❀1
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ†• StreamSpeech: A powerful synchronized speech translation model.

StreamSpeech is a seamless All-in-One model for offline and synchronous speech recognition, speech translation and speech synthesis.

πŸ’‘ StreamSpeech achieves SOTA performance for both offline and synchronous speech-to-speech translation.

🟒page: https://ictnlp.github.io/StreamSpeech-site/

🟒paper: https://arxiv.org/abs/2406.03049

🟒code: https://github.com/ictnlp/streamspeech

https://t.me/DataScienceT βœ…
Please open Telegram to view this post
VIEW IN TELEGRAM
πŸ‘3
⚑ DeepSeek-Coder-V2: The first open source model to outperform GPT4-Turbo in coding and math problems

> > > Outperforms GPT4-Turbo, Claude3-Opus, Gemini-1.5Pro, Codestral in coding and math problems.
> Supports 338 programming languages, 128K context length.
> Fully open source code in two sizes: 230B and 16 B

DeepSeek-Coder-V2 outperforms Yi-large, Claude3-Opus, GL M4 and Qwen2-72B in the Arena-Hard-Auto table.

β–ͺ️HF: https://huggingface.co/deepseek-ai/DeepSeek-Coder-V2-Instruct

β–ͺ️Github: https://github.com/deepseek-ai/DeepSeek-Coder-V2/blob/main/paper.pdf

β–ͺ️Demo: https://chat.deepseek.com/sign_in?from=coder

https://t.me/DataScienceT βœ…
Please open Telegram to view this post
VIEW IN TELEGRAM
πŸ‘4
TSI-Bench: Benchmarking Time Series Imputation

πŸ–₯ Github: https://github.com/WenjieDu/Awesome_Imputation

πŸ“• Paper: https://arxiv.org/pdf/2406.12747v1.pdf

πŸ’™ Dataset: https://github.com/WenjieDu/TSDB

https://t.me/DataScienceT βœ…
Please open Telegram to view this post
VIEW IN TELEGRAM
πŸ‘3