Igor Melnyk
Arxiv Papers
Running out of time to catch up with new arXiv papers? We take the most impactful papers and present them as convenient podcasts. If you're a visual learner, we offer these papers in an engaging video format. Our service fills the gap between overly brief paper summaries and time-consuming full paper reads. You gain academic insights in a time-efficient, digestible format. Code behind this work: https://github.com/imelnyk/ArxivPapers
Where to listen?
Podcasts in the app Replaio Radio Coming soonPodcasts are coming to the app soon. Install now and be the first to see a whole new take on podcasts
Install for free
Android 5M+ downloads · 4.8 rating
iOS soon
Episodes
Think Big, Generate Quick: LLM-to-SLM for Fast Autoregressive Decoding 27.02.2024 28:10
Hybrid approach combines large and small language models for efficient autoregressive decoding, achieving speedups of up to 3x with minor performance trade-offs. https://arxiv.org/abs//2402.16844 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016 Spotify: https://podcasters.spoti...
[short] Watermarking Makes Language Models Radioactive 26.02.2024 2:54
https://arxiv.org/abs//2402.14904 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016 Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers
Watermarking Makes Language Models Radioactive 26.02.2024 26:23
https://arxiv.org/abs//2402.14904 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016 Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers
[short] Same Task, More Tokens: the Impact of Input Length on the Reasoning Performance of Large Language Models 26.02.2024 2:47
The paper examines the impact of input length extension on Large Language Models, revealing a notable decrease in reasoning performance at shorter lengths than expected, with insights for future research. https://arxiv.org/abs//2402.14848 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1...
Same Task, More Tokens: the Impact of Input Length on the Reasoning Performance of Large Language Models 26.02.2024 22:17
The paper examines the impact of input length extension on Large Language Models, revealing a notable decrease in reasoning performance at shorter lengths than expected, with insights for future research. https://arxiv.org/abs//2402.14848 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1...
[short] Gen4Gen: Generative Data Pipeline for Generative Multi-Concept Composition 26.02.2024 2:32
The paper introduces Gen4Gen, a dataset creation pipeline for multi-concept personalization in text-to-image models, with a new metric and baseline for evaluation, improving image generation quality. https://arxiv.org/abs//2402.15504 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id169247...
Gen4Gen: Generative Data Pipeline for Generative Multi-Concept Composition 26.02.2024 17:00
The paper introduces Gen4Gen, a dataset creation pipeline for multi-concept personalization in text-to-image models, with a new metric and baseline for evaluation, improving image generation quality. https://arxiv.org/abs//2402.15504 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id169247...
[short] LongRoPE: Extending LLM Context Window Beyond 2 Million Tokens 25.02.2024 2:25
LongRoPE extends context window of LLMs to 2048k tokens efficiently, maintaining performance, with innovative strategies and minor modifications, demonstrated effective across tasks. https://arxiv.org/abs//2402.13753 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016 Spotify: htt...
[short] LongRoPE: Extending LLM Context Window Beyond 2 Million Tokens 25.02.2024 28:02
LongRoPE extends context window of LLMs to 2048k tokens efficiently, maintaining performance, with innovative strategies and minor modifications, demonstrated effective across tasks. https://arxiv.org/abs//2402.13753 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016 Spotify: htt...
[short] Beyond A*: Better Planning with Transformers via Search Dynamics Bootstrapping 24.02.2024 2:11
https://arxiv.org/abs//2402.14083 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016 Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers
Beyond A*: Better Planning with Transformers via Search Dynamics Bootstrapping 24.02.2024 27:23
https://arxiv.org/abs//2402.14083 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016 Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers
[short] OmniPred: Language Models as Universal Regressors 23.02.2024 2:40
OMNIPRED proposes a universal regression framework using language models trained on diverse experimental data, outperforming traditional regression models by solely using textual representations of parameters. https://arxiv.org/abs//2402.14547 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-paper...
OmniPred: Language Models as Universal Regressors 23.02.2024 17:13
OMNIPRED proposes a universal regression framework using language models trained on diverse experimental data, outperforming traditional regression models by solely using textual representations of parameters. https://arxiv.org/abs//2402.14547 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-paper...
[short] T-Stitch: Accelerating Sampling in Pre-Trained Diffusion Models with Trajectory Stitching 23.02.2024 2:58
The paper introduces T-Stitch, a technique for efficient sampling from diffusion probabilistic models by using a smaller model initially and switching to a larger model later, improving efficiency without compromising quality. https://arxiv.org/abs//2402.14167 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/po...
T-Stitch: Accelerating Sampling in Pre-Trained Diffusion Models with Trajectory Stitching 23.02.2024 18:46
The paper introduces T-Stitch, a technique for efficient sampling from diffusion probabilistic models by using a smaller model initially and switching to a larger model later, improving efficiency without compromising quality. https://arxiv.org/abs//2402.14167 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/po...
[short] Back to Basics: Revisiting REINFORCE Style Optimization for Learning from Human Feedback in LLMs 23.02.2024 2:15
https://arxiv.org/abs//2402.14740 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016 Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers
Back to Basics: Revisiting REINFORCE Style Optimization for Learning from Human Feedback in LLMs 23.02.2024 25:32
https://arxiv.org/abs//2402.14740 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016 Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers
[short] In deep reinforcement learning, a pruned network is a good network 22.02.2024 2:33
Gradual magnitude pruning enhances deep reinforcement learning agents' parameter effectiveness, leading to significant performance gains and adherence to a "scaling law" with minimal network parameters. https://arxiv.org/abs//2402.12479 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxi...
In deep reinforcement learning, a pruned network is a good network 22.02.2024 16:45
Gradual magnitude pruning enhances deep reinforcement learning agents' parameter effectiveness, leading to significant performance gains and adherence to a "scaling law" with minimal network parameters. https://arxiv.org/abs//2402.12479 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxi...
[short] Coercing LLMs to do and reveal (almost) anything 22.02.2024 2:16
Adversarial attacks on large language models extend beyond jailbreaking, encompassing misdirection, model control, denial-of-service, and data extraction. Comprehensive security measures are crucial. https://arxiv.org/abs//2402.14020 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id169247...
Coercing LLMs to do and reveal (almost) anything 22.02.2024 26:49
Adversarial attacks on large language models extend beyond jailbreaking, encompassing misdirection, model control, denial-of-service, and data extraction. Comprehensive security measures are crucial. https://arxiv.org/abs//2402.14020 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id169247...
[short] Synthetic Data (Almost) from Scratch: Generalized Instruction Tuning for Language Models 21.02.2024 2:33
GLAN introduces a scalable method for instruction tuning of Large Language Models using a pre-curated taxonomy of human knowledge, generating diverse instructions across disciplines. https://arxiv.org/abs//2402.13064 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016 Spotify: htt...
Synthetic Data (Almost) from Scratch: Generalized Instruction Tuning for Language Models 21.02.2024 26:57
GLAN introduces a scalable method for instruction tuning of Large Language Models using a pre-curated taxonomy of human knowledge, generating diverse instructions across disciplines. https://arxiv.org/abs//2402.13064 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016 Spotify: htt...
[short] Neural Network Diffusion 21.02.2024 2:13
Diffusion models can generate high-performing neural network parameters by utilizing an autoencoder and latent diffusion model, showing comparable or improved performance over trained networks. https://arxiv.org/abs//2402.13144 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016 S...
Neural Network Diffusion 21.02.2024 9:26
Diffusion models can generate high-performing neural network parameters by utilizing an autoencoder and latent diffusion model, showing comparable or improved performance over trained networks. https://arxiv.org/abs//2402.13144 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016 S...
Similar podcasts
Listen to the Arxiv Papers podcast in Replaio
Radio and podcasts in one app - free, with no sign-up. Install today and do not miss the launch
Replaio is not a podcast publisher; show names, artwork and audio belong to their authors and are distributed through public RSS feeds.
