phi_equation

"HSTU-based Generative Recommenders, with 1.5 trillion parameters, improve metrics in online A/B tests by 12.4% and have been deployed on multiple surfaces of a large internet platform with billions of users. More importantly, the model quality of Generative Recommenders empirically scales as a power-law of training compute across three orders of magnitude, up to GPT-3/LLaMa-2 scale."

"we train a series of embodied diffusion models and align them into the LLM for predicting the goal images and point clouds"

"we also use a self-supervised loss that focuses on modeling inter-sentence coherence"

"Our theoretical study shows that the upper bound of interpolation is at least ∼600× smaller than that of extrapolation"

"Another application of this procedure is a polynomial quantum Fourier transform algorithm for an arbitrary finite Abelian group."

"These capabilities enable more reliable synthetic data generation, policy evaluation, and closed-loop simulation for robotics and autonomous systems"

"G-EVAL is a prompt-based evaluator with three main components: 1) a prompt that contains the definition of the evaluation task and the desired evaluation criteria, 2) a chain-of-thoughts (CoT) ..."

"Instructions for using the evaluation server are provided."

"MMBench is a systematically designed objective benchmark for a robust and holistic evaluation of vision-language models"

"For all significant gravitational-wave events, we provide estimates of the source properties... using relativistic models of GWs from CBCs"

"We use CLIP to compute embeddings of the image and alt-text. Then we compute the cosine similarity of both embeddings and drop all samples with cosine similarity below 0.3"

"DeepSeek-OCR consists of two components: DeepEncoder and DeepSeek3B-MoE-A570M as the decoder."

"Large but sparsely activated DNNs can consume <1/10th the energy of large, dense DNNs without sacrificing accuracy despite using as many or even more parameters"

"Gorilla's retriever–aware training enables it to react to changes in the APIs."

"outperforms baseline methods by up to 181% across... HPS-v2.1, CLIP Score, VideoAlign, and GenEval"

"RetNet achieves favorable scaling results, parallel training, low-cost deployment, and efficient inference"

"The DeepSeek-VL family (both 1.3B and 7B models) showcases superior user experiences as a vision-language chatbot in real-world applications, achieving state-of-the-art or competitive performance across a wide range of visual-language benchmarks at the same model size while maintaining robust performance on language-centric benchmarks."

"We present GLM-4.5, an open-source Mixture-of-Experts (MoE) large language model with 355B total parameters and 32B activated parameters, featuring a hybrid reasoning method that supports both thinking and direct response modes."

"decoupling visual encoding for multimodal understanding and generation"

"Guided by the scaling laws, we introduce DeepSeek LLM, a project dedicated to advancing open-source language models with a long-term perspective."

"Planck temperature, polarization, and lensing data determine the spectral index of scalar perturbations to be ns=0.9649±0.0042 at 68% CL and show no evidence for a scale dependence of ns."

"We fine-tune and evaluate our models on a wide range of knowledge-intensive NLP tasks and set the state-of-the-art on three open domain QA tasks, outperforming parametric seq2seq models and task-specific retrieve-and-extract architectures."

"phi-3-mini achieves 69% on MMLU and 8.38 on MT-bench, rivaling Mixtral 8x7B and GPT-3.5"

"we have evaluated 18 base LLMs and 34 instruction-tuned LLMs on LiveCodeBench"

"identify a roughly linear relation between the RL reward and the square root of the KL divergence"

"our high-compute variant, DeepSeek-V3.2-Speciale, surpasses GPT-5 and exhibits reasoning proficiency on par with Gemini-3.0-Pro, achieving gold-medal performance in both the 2025 International Mathematical Olympiad (IMO) and the International Olympiad in Informatics (IOI)"

"We propose a new test to measure a text model's multitask accuracy. The test covers 57 tasks including elementary mathematics, US history, computer science, law, and more."

"achieving greater than 1000x improvements in model capacity ... a MoE with up to 137 billion parameters is applied convolutionally between stacked LSTM layers."

"We hypothesize that the change in weights during model adaptation also has a low 'intrinsic rank'... W0 + ΔW = W0 + BA, where B∈ℝ^{d×r}, A∈ℝ^{r×k}, and the rank r ≪ min(d,k)."

"PE(pos,2i) = sin(pos/10000^{2i/d_model}); PE(pos,2i+1) = cos(pos/10000^{2i/d_model})"