Arxiv Papers

The paper criticizes the Fréchet Inception Distance (FID) as an evaluation metric for generated images and proposes an alternative metric called CMMD, which is based on CLIP embeddings and offers a more reliable assessment of image quality. https://arxiv.org/abs//2401.09603 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcasts.a...

The paper introduces ChatQA, a conversational question answering model that achieves GPT-4 level accuracies. It proposes a two-stage instruction tuning method and a dense retriever for retrieval in conversational QA. ChatQA-70B outperforms GPT-4 without using synthetic data. https://arxiv.org/abs//2401.10225 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers...

The paper introduces ChatQA, a conversational question answering model that achieves GPT-4 level accuracies. It proposes a two-stage instruction tuning method and a dense retriever for retrieval in conversational QA. ChatQA-70B outperforms GPT-4 without using synthetic data. https://arxiv.org/abs//2401.10225 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers...

The paper discusses the use of machine learning algorithms to predict the risk of cardiovascular disease based on electronic health records. https://arxiv.org/abs//2401.09865 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...

The paper discusses the use of machine learning algorithms to predict the risk of cardiovascular disease based on electronic health records. https://arxiv.org/abs//2401.09865 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...

To achieve superhuman agents, future models need superhuman feedback. This paper proposes Self-Rewarding Language Models, where the model provides its own rewards during training, leading to improved performance and the potential for continual improvement. https://arxiv.org/abs//2401.10020 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: ht...

To achieve superhuman agents, future models need superhuman feedback. This paper proposes Self-Rewarding Language Models, where the model provides its own rewards during training, leading to improved performance and the potential for continual improvement. https://arxiv.org/abs//2401.10020 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: ht...

The paper proposes a method called Reinforced Fine-Tuning (ReFT) to enhance the generalizability of Large Language Models (LLMs) for reasoning tasks, using math problem-solving as an example. ReFT combines Supervised Fine-Tuning (SFT) with reinforcement learning and outperforms SFT in experiments. https://arxiv.org/abs//2401.08967 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.t...

The paper proposes a method called Reinforced Fine-Tuning (ReFT) to enhance the generalizability of Large Language Models (LLMs) for reasoning tasks, using math problem-solving as an example. ReFT combines Supervised Fine-Tuning (SFT) with reinforcement learning and outperforms SFT in experiments. https://arxiv.org/abs//2401.08967 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.t...

The paper discusses the use of machine learning algorithms to predict the risk of cardiovascular disease based on electronic health records. https://arxiv.org/abs//2401.09135 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...

The paper discusses the use of machine learning algorithms to predict the risk of cardiovascular disease based on electronic health records. https://arxiv.org/abs//2401.09135 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...

DeepSpeed-FastGen is a system that uses Dynamic SplitFuse to improve the deployment and scaling of large language models, achieving higher throughput and lower latency compared to existing systems. The code is available for community engagement. https://arxiv.org/abs//2401.08671 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podca...

DeepSpeed-FastGen is a system that uses Dynamic SplitFuse to improve the deployment and scaling of large language models, achieving higher throughput and lower latency compared to existing systems. The code is available for community engagement. https://arxiv.org/abs//2401.08671 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podca...

Proxy-tuning is a lightweight decoding-time algorithm that can be used to customize large pretrained language models without accessing their weights. It achieves similar results to direct tuning and can be applied for domain adaptation and task-specific finetuning. https://arxiv.org/abs//2401.08565 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Pod...

Proxy-tuning is a lightweight decoding-time algorithm that can be used to customize large pretrained language models without accessing their weights. It achieves similar results to direct tuning and can be applied for domain adaptation and task-specific finetuning. https://arxiv.org/abs//2401.08565 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Pod...

This paper introduces AIM, a collection of vision models pre-trained with an autoregressive objective. The models exhibit similar scaling properties to Large Language Models (LLMs) and achieve high performance on downstream tasks. Pre-training AIM does not require image-specific strategies. https://arxiv.org/abs//2401.08541 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.c...

This paper introduces AIM, a collection of vision models pre-trained with an autoregressive objective. The models exhibit similar scaling properties to Large Language Models (LLMs) and achieve high performance on downstream tasks. Pre-training AIM does not require image-specific strategies. https://arxiv.org/abs//2401.08541 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.c...

The paper challenges the belief that the area under the precision-recall curve (AUPRC) is a superior metric for model comparison in machine learning. It argues that AUPRC can be biased and highlights a lack of empirical evidence supporting its supposed advantages. https://arxiv.org/abs//2401.06091 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podc...

The paper challenges the belief that the area under the precision-recall curve (AUPRC) is a superior metric for model comparison in machine learning. It argues that AUPRC can be biased and highlights a lack of empirical evidence supporting its supposed advantages. https://arxiv.org/abs//2401.06091 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podc...

The paper discusses the use of machine learning algorithms to predict the risk of cardiovascular disease in patients with diabetes. https://arxiv.org/abs//2312.00886 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

The paper discusses the use of machine learning algorithms to predict the risk of cardiovascular disease in patients with diabetes. https://arxiv.org/abs//2312.00886 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

Current language models often generalize well from easy to hard data, performing as well as models trained on hard data. It may be better to collect and train on easy data rather than hard data, as hard data is noisier and costlier to collect. https://arxiv.org/abs//2401.06751 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcast...

Current language models often generalize well from easy to hard data, performing as well as models trained on hard data. It may be better to collect and train on easy data rather than hard data, as hard data is noisier and costlier to collect. https://arxiv.org/abs//2401.06751 YouTube: https://www.youtube.com/@ArxivPapers TikTok: https://www.tiktok.com/@arxiv_papers Apple Podcasts: https://podcast...

The paper examines the data curation process for large language models and investigates how different filters affect webpages based on social and geographic dimensions. The study highlights implicit preferences in data curation and calls for further research on the social implications of pretraining data curation. https://arxiv.org/abs//2401.06408 YouTube: https://www.youtube.com/@ArxivPapers TikT...

The paper examines the data curation process for large language models and investigates how different filters affect webpages based on social and geographic dimensions. The study highlights implicit preferences in data curation and calls for further research on the social implications of pretraining data curation. https://arxiv.org/abs//2401.06408 YouTube: https://www.youtube.com/@ArxivPapers TikT...