Researcher profile

Gang Xie

Gang Xie contributes to research discovery and scholarly infrastructure.

ResearcherAffiliation not importedOpen to collaborate

hep-ph Artificial Intelligence Computation and Language

Trust snapshot

Quick read

Trust 17 - Baseline

4works

0followers

3topics

4close collaborators

Actions

Decide how to stay connected

Follow researcher0

Inspect adjacent work, topics, institutions and collaborators without jumping out to a separate graph page.

Building this graph slice

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2026arXiv

MiMo-V2-Flash Technical Report

We present MiMo-V2-Flash, a Mixture-of-Experts (MoE) model with 309B total parameters and 15B active parameters, designed for fast, strong reasoning and agentic capabilities. MiMo-V2-Flash adopts a hybrid attention architecture that interleaves Sliding Window Attention (SWA) with global attention, with a 128-token sliding window under a 5:1 hybrid ratio. The model is pre-trained on 27 trillion tokens with Multi-Token Prediction (MTP), employing a native 32k context length and subsequently extended to 256k. To efficiently scale post-training compute, MiMo-V2-Flash introduces a novel Multi-Teacher On-Policy Distillation (MOPD) paradigm. In this framework, domain-specialized teachers (e.g., trained via large-scale reinforcement learning) provide dense and token-level reward, enabling the student model to perfectly master teacher expertise. MiMo-V2-Flash rivals top-tier open-weight models such as DeepSeek-V3.2 and Kimi-K2, despite using only 1/2 and 1/3 of their total parameters, respectively. During inference, by repurposing MTP as a draft model for speculative decoding, MiMo-V2-Flash achieves up to 3.6 acceptance length and 2.6x decoding speedup with three MTP layers. We open-source both the model weights and the three-layer MTP weights to foster open research and community collaboration.

preprint2022arXiv

Exclusive vector meson productions with the analytical solution of Balitsky-Kovchegov Equation

Exclusive vector meson production is an excellent probe for describing the structure of proton. In this paper, based on dipole model, the differential cross sections, total cross sections and the ratios of the longitudinal to transverse cross sections of $J/ψ$ and $ρ^0$ productions are calculated with the analytical solution of Balitsky-Kovchegov (BK) equation. In addition, we also consider the influences of two meson wave function models on the results. Our predictions, which are little sensitive to meson wave functions, agree with the experimental data. The analytical solution of BK equation is reliable for description of exclusive vector meson production in a certain range of $Q^2$.

preprint2022arXiv

The neutron and proton mass radii from the vector meson photoproduction data on the deuterium target

In this study, we try to extract the mass radii of the neutron and the proton from the differential cross section data of near-threshold $ω$ and $ϕ$ photoproductions on deuterium target, which is often approximated as a quasi-free neutron plus a quasi-free proton. The incoherent data of $ω$ and $ϕ$ photoproductions are provided by CBELSA/TAPS collaboration and LEPS collaboration respectively, where the deuteron is disintegrated in the experiments to measure the properties of individual nucleons. Under the VMD model and the assumption of dipole gravitational form factor, we determined the loosely bound neutron and proton mass radii to be $0.795\pm0.092\rm(stat.)\pm0.073\rm(syst.)$ fm and $0.744\pm0.029\rm(stat.)\pm0.042\rm(syst.)$ fm respectively from the near-threshold data of $γd \rightarrow ωn (p)$ and $γd \rightarrow ωp (n)$, for the first time. With the near-threshold and incoherent $ϕ$ photoproduction data of $γd \rightarrow ϕp n$, we determined the average mass radius of the bound nucleon (neutron or proton) inside the deuteron to be $0.755\pm0.039\rm(stat.)\pm0.039\rm(syst.)$ fm, for the first time. For a comparison study, we also extracted the mass radius of the free proton from the $ω$ photoproduction on the hydrogen target by CBELSA/TAPS collaboration. Based on our analysis results under the assumptions of VMD model and a low energy QCD theorem, we find that the neutron mass radius is consistent with the proton mass radius within the current statistical uncertainties, and that the nuclear modification on the nucleon mass radius is small inside the deuteron.

preprint2021arXiv

Tackling the kaon structure function at EicC

Measuring the kaon structure beyond the proton and pion structures is one of the hot topics in hadron physics, as it is one way to understand the nature of Nambu-Goldstone boson of QCD and to see the interplay between the EHM mechanism and the HB mechanism for hadron mass generation. In this paper, we present a simulation of the leading $Λ$ baryon tagged deep inelastic scattering experiment at EicC (Electron-ion collider in China), which is engaged to unveil the internal structure of kaon through Sullivan process. According to our simulation results, the suggested experiment will cover the kinematical domain of $0.05\lesssim x_{\rm K} \lesssim 0.85$ and $Q^2$ up to 50 GeV$^2$, with the acceptable statistical uncertainties. In the relatively low-$Q^2$ region ($<10$ GeV$^2$), the Monte-Carlo simulation shows a good precision of the measurement ($<5$\%) for the kaon structure function $F_2^{\rm K}$. In the high-$Q^2$ region (up to 50 GeV$^2$), the statistical uncertainty of $F_2^{\rm K}$ is also acceptable ($<10$\%) for the data at $x_{\rm K}<0.8$. To perform such an experiment at an electron-ion collider, a high-performance zero-degree calorimeter is required.