algorithms

New Algorithms and Approaches

hallucinationsdiffusionvisionagentssecurity-compliancetranslationpromptingsafetydeepfakes & fake newsmusicarchitectureRAGalgorithmsfinancevoice and text-to-speechmodelsvideo and text-to-videohealthcaredefense
diffusiondiffusion
+5+5
[#30] Meta releases better technique for more efficient video-language understanding

[#30] Meta releases better technique for more efficient video-language understanding

Plus AMD reduces the cost of training diffusion models, lung cancer detection, autonomous driving from aerial photos, and seamless AI voice conversation

Chris Han
Chris Han
hallucinationshallucinations
+3+3
[#24] Meta releases Llama 3.2, models for mobile & edge devices

[#24] Meta releases Llama 3.2, models for mobile & edge devices

Plus detecting cheating during exams with AI, AI2 releases open-source language models, mitigating hallucinations in vision language models, and zero-shot detection of AI images

Chris Han
Chris Han
visionvision
+4+4
[#23] Apple's HyperCloning: Training Using Small Models

[#23] Apple's HyperCloning: Training Using Small Models

Plus Amazon innovates on product translations, Anthropic releases Contextual Retrieval, Detecting deepfakes with smartphones, and RLHF can mislead humans

Chris Han
Chris Han
translationtranslation
+4+4
[#21] Nvidia releases new SOTA reranking model

[#21] Nvidia releases new SOTA reranking model

Plus Microsoft releases a model to simulate financial markets, Meta researchers release Source2Synth, an LLM that can predict stock prices, and Google finds a better way to translate long-text with LLMs

Chris Han
Chris Han
security-compliancesecurity-compliance
+3+3
[#20] Are better deepfakes here? TikTok's parent company unveils enhanced dubbing framework.

[#20] Are better deepfakes here? TikTok's parent company unveils enhanced dubbing framework.

Plus research from DeepMind, indecision can be incorporated into recommender systems, a dataset for jailbreaking LLMs, text-to-3D models,

Chris Han
Chris Han
architecturearchitecture
+1+1
[#19] ScaleAI improves how LLMs generate code

[#19] ScaleAI improves how LLMs generate code

Plus predicting if a startup will be successful, how TikTok content can be moderated, low-light image enhancement, xAI, and a new 7B parameter model from AI2

Chris Han
Chris Han
agentsagents
+3+3
[#18] How to detect LLM-generated code

[#18] How to detect LLM-generated code

Plus PatentGPT can generate patents automatically, NVIDIA defends RAG vs long-context windows, facial recognition on pixelated images, and why AI agents perform poorly

Chris Han
Chris Han
architecturearchitecture
+2+2
[#17] Deepfake video detection

[#17] Deepfake video detection

Plus warships designed by AI, combating model collapse, and synthetic data with small LLMs

Chris Han
Chris Han
security-compliancesecurity-compliance
+3+3
[#15] Meta releases Human Vision Models

[#15] Meta releases Human Vision Models

Plus new models from A21Labs, generating sketches from an image, a database of AI risks, and answering questions about a first person scene

Chris Han
Chris Han
agentsagents
+2+2
[#14] How to edit the clothes on fashion models

[#14] How to edit the clothes on fashion models

Plus long-form video generation with agents, quantum computing and AI, LLMs actually generate insecure code, and models that can detect camouflaged data

Chris Han
Chris Han
architecturearchitecture
+2+2
Issue #13: Amazon creates new RAG evaluation framework

Issue #13: Amazon creates new RAG evaluation framework

Plus automated scientific discovery, improving data quality, flexible and visual memory, predicting future events

Chris Han
Chris Han
security-compliancesecurity-compliance
+3+3
Issue #11: How Meta recommends short form videos

Issue #11: How Meta recommends short form videos

Plus Snowflake's Arctic-TILT LLM architecture, achieving better citations than ChatGPT, stopping adversarial AI attacks at the edge, and achieving better alignment with LLMs

Chris Han
Chris Han