搜索结果：H-1B

The Fundamental Review of the Trading Book (FRTB) poses a significant challenge for exotic derivatives pricing, particularly for non-modelable risk factors (NMRF) where sparse market data leads to infinite audit bounds under classical Martingale Optimal Transport (MOT). We propose a unified Rough Martingale Optimal Transport (RMOT) framework that regularizes the transport plan with a rough volatility prior, yielding finite, explicit, and asymptotically tight extrapolation bounds. We establish an identifiability theorem for rough volatility parameters under sparse data, proving that 50 strikes are sufficient to estimate the Hurst exponent within $\pm 0.05$. For the multi-asset case, we prove that the correlation matrix is locally identifiable from marginal option surfaces provided the Hurst exponents are distinct. Model calibration on SPY and QQQ options (2019--2024) confirms that the optimal martingale measure exhibits stretched exponential tail decay ($\sim\exp(-k^{1-H})$), consistent with rough volatility asymptotics, whereas classical MOT yields trivial bounds. We validate the framework on live SPX/NDX data and scale it to $N = 30$ assets using a block-sparse optimization algori

Plasma composition in the solar corona commonly differs from that of the photosphere, with the enhancement of low--first-ionization-potential (FIP) elements referred to as the FIP effect. This phenomenon provides important diagnostics of energy and mass transport between different layers of the solar atmosphere. In this work, we analyze an anomalously strong, localized FIP effect observed in active region 13486 associated with a subflaring episode on 2023 November 17, using multiwavelength observations combining high energy-resolution soft X-ray disk-integrated spectra obtained by the Macao Science Satellite-1B with spatially resolved EUV/UV and H$α$ imaging from Hinode/EIS, SDO/AIA and HMI, and CHASE/HIS. By investigating the temporal evolution of plasma composition in response to changes in magnetic field orientation, we provide new insight into the physical processes linking magnetic reconnection, ponderomotive force fractionation, and coronal abundance anomalies. This work reveals that the anomalously strong enhancement of low-FIP elements is localized in regions with strongly inclined magnetic fields despite a subflare. We interpret these observations within the framework of t

IndicTrans2 is the strongest open English to Indic translation system, but like most systems it is trained on general text and tends to sound stiff on casual, conversational input. We adapt IndicTrans2-1B to conversational register across all 21 Indic languages using only public data (OpenSubtitles, BPCC-H-Daily, Tatoeba). Plain fine-tuning improves conversational chrF but forgets the general domain (it drops 3.9 chrF on FLORES for Hindi). Mixing general data back into training (experience replay) and then averaging the fine-tuned weights with the base (model souping) removes that trade-off: the resulting model beats IndicTrans2-1B on conversational chrF in every one of the 21 languages (mean +6.2) while matching it on FLORES (mean change -0.17, all within 0.7 chrF). Paired bootstrap tests confirm the conversational gains are significant (p <= 0.004) and that FLORES is not significantly degraded. We are deliberate about scope: these are chrF gains, and a blind human plus multi-model LLM check does not confirm them as a perceived quality improvement, so we treat the conversational gain as largely a register match to the references rather than proof of better translation. The tech

Available JWST observations TRAPPIST-1 system have suggested that several of the planets are likely airless, or possess a very tenuous atmosphere. However, the high atmospheric escape rates expected for these planets suggest that any tenuous atmosphere must be replenished by constant outgassing, and past studies on modeling potential atmospheres for the planets have not widely considered surface pressures <1 bar. Here, we show that tenuous atmospheres on the TRAPPIST-1 planets are likely possible, supported by constant plausible rates of water and/or CO$_{2}$ outgassing against assumed high escape rates (up to ~10$^{30}$ s$^{-1}$). We use a coupled photochemical-climate model and sample from a broad phase space of outgassing, surface deposition, and top-of-atmosphere escape rates to test hundreds of atmospheres per planet. Critically, our model also allows surface pressure to vary based on the balance of sources and sinks. We find that 6 different compositional archetypes are generated via H$_{2}$O and/or CO$_{2}$ outgassing across our phase space, and atmospheres commonly fall between 10$^{-4}$ -- 1 bar. We find that potentially habitable surface environments are possible for T

Frontier language capability is usually bought with frontier compute; CHERRY shows a different trade. It is a sovereign Korean model family built on one principle: supervise the tokens that decide the answer, and let shared weights carry the rest. Under matched compute this exposes a sharp, reproducible dissociation---selected-token supervision preserves held-out discrimination yet collapses free generation, and a full-sequence anchor recovers only part of the gap. The same signal drives a heal-after-merge recurrent-representational-yield loop that collapses 48 layers to 6 unique blocks at near-dense parity (227M at loss 2.934 vs a 566M dense model at 2.926) and composes them by MoEE fusion (2.789)---a recurrent-compression direction independently pursued by concurrent frontier looped-MoE work, which we project (not yet measure) to frontier scale. It also installs metacognition from two-token supervision (200 held-out KO prompts/type, kappa>0.82, +/-6.9pp): self-correction 12->47% and jailbreak 23->4% at 97.6% loss-retention on 1.2B, with a pre-registered 1B->13.7B ablation localising the operand-binding limit to capacity (1B lookup vs 13.7B H-PRESERVE); and it speciali

We present new Keck/KPIC high-resolution spectroscopic detections of three ultra-hot Jupiters (UHJs) in the $K$ band: WASP-189b ($\rm SNR = 7.2$), MASCARA-1b ($\rm SNR = 8.6$), and TOI-1518b ($\rm SNR = 7.1$), as well as a tentative detection of KELT-9b ($\rm SNR = 5.0$). We perform a uniform set of atmospheric retrieval analysis on these objects, as well as previously reported KPIC observations of WASP-33b ($\rm SNR = 11.2$) and KELT-20b ($\rm SNR = 10.5$), We perform atmospheric retrievals for the pressure-temperature ($P-T$) profile, orbital velocity parameters, $v\sin i$, and abundances of CO, H$_2$O, OH, and Fe, with parameterized mixing profiles to account for the expected vertical abundance variations of H$_2$O and OH. We also perform a set of retrievals assuming chemical equilibrium, which are generally in good agreement with the free retrievals. Except for \knb, the retrieved spectra are dominated by CO emission features, with additional weak H$_2$O or OH features consistent with thermal dissociation of H$_2$O. \knb, which is significantly hotter, appears to have very weak molecular features. Dissociation limits our ability to reliably constrain H$_2$O or OH abundances fro

We present Harmonic, a hierarchical state space model (SSM) for language modeling. The architecture stacks three recurrent levels at progressively slower timescales; each level receives the prediction error of the level below as input, rather than its raw hidden state. On enwiki8 with equal token budgets, Harmonic outperforms a comparable Transformer (28M params) by +1.4% at 1K tokens, +6.7% at 8K tokens, and +11.4% at 32K tokens (bpt, lower is better). It also outperforms Mamba at every tested length by 0.7--1.8%. At 64K tokens, both Mamba and Transformer run out of memory on an 80GB H100; Harmonic trains successfully, reaching 6.169 bpt. Results replicate on WikiText-103 (H-TF gap +1.7% to +7.2% across 1K--32K). At 1B parameter scale, replacing all attention layers in TinyLlama 1.1B with HarmonicBlock eliminates the RoPE positional encoding limit: the resulting Hallamonic model maintains stable loss across sequence lengths 1K--8K on two independent clean benchmarks (Lambada and fineweb-edu held-out), while TinyLlama degrades catastrophically past its 2K-token RoPE limit (gap: +9.4 bpt at seq=8K on Lambada). Compute is O(L) per forward pass vs. O(L^2) for attention. Logs: https://

Ultra-hot Jupiters (UHJs; $T_{\rm eq} \gtrsim 2000$ K) enable simultaneous detection of volatile (ice-forming) and refractory (rock-forming) species in planetary atmospheres, providing a powerful diagnostic of planet formation and atmospheric processing. We present a comprehensive high-resolution cross-correlation spectroscopy (HRCCS) analysis of the UHJ MASCARA-1b ($T_{\rm eq} \approx 2600$ K) using the IGRINS and IGRINS-2 spectrographs. We detect robust (SNR$>$4) signals from H$_2$O, CO, OH, Fe I, Mg I, Ca I, and Ti I, marking the most complete atmospheric inventory of MASCARA-1b to date. Using a chemically consistent atmospheric inference framework, we constrain elemental abundances to a typical precision of $\approx$0.2 dex, retrieving a solar atmospheric metallicity ([M/H]$_\odot$ $= 0.07^{+0.17}_{-0.13}$ $\approx 1.2\times$ solar), a C/O ratio (C/O $= 0.65^{+0.08}_{-0.08}$) consistent with solar value (C/O $=$ 0.59), an enhanced refractory abundance ([R/H]$_\odot$ $= 0.40^{+0.23}_{-0.17} \approx 2.5\times$ solar; $\approx 3.8\times$ stellar), and a moderately super-solar refractory-to-volatile ratio ([R/V]$_\odot$ $= 0.36^{+0.11}_{-0.09}$ $\approx 2.3\times$ solar). Compar

Transformers propagate information across depth through a single additive residual stream: every sublayer reads only the most recent state. Attention residuals relax this by letting each sublayer attend, through a learned softmax. However, that read uses a single query shared across the entire width, so every feature subspace must read the depth history through one distribution. The cost of this forced compromise grows with how much the subspaces disagree about which layers to read, and disagreement grows with model width. We introduce Multi-Head Attention Residuals (MHAR): the routing query is reshaped into H per-subspace heads, each with its own softmax over the depth history. The read becomes block-diagonal, the reshape adds zero parameters and negligible compute, and H = 1 recovers attention residuals exactly. Trained from scratch on a deduplicated Nemotron-based anneal corpus that is quality-filtered and STEM- and code-heavy, MHAR improves validation loss over a standard Transformer at 100M, 350M, and 1B (-0.061, -0.149, and -0.140). It achieves the best result among four methods in every setting, with the gain increasing from 100M to the larger scales. The head count is a rea

We present the first discoveries from Keck Observations in the INfrared of Taurus and $ρ$ Oph Exoplanets And Ultracool dwarfs (KOINTREAU), an adaptive optics imaging survey of young stars in the Taurus and $ρ$ Oph star-forming regions using the Keck infrared pyramid wavefront sensor (PyWFS). We have found two faint ($Δ$K~7 mag), wide-separation companions to two ~3-Myr-old Taurus members. Relative astrometry for these systems show that both companions are bound to their host stars. We obtained near-infrared spectra of these companions using IRTF/SpeX (R~100) and Gemini/GNIRS (R~1000-2000), and combine these with photometry from our NIRC2 imaging, the Pan-STARRS survey, and Spitzer/IRAC archival imaging to constrain their properties. One companion, KOINTREAU-1b (at a projected separation of 690 au), has an average near-IR spectral type of M9$\pm$2, a gravity classification of VL-G, and a changing spectral type between the SpeX (M7) and GNIRS (L1) observations. We estimate this object's mass to be $10.6^{+2.5}_{-2.3}$ M$_{\rm Jup}$, making KOINTREAU-1b the fifth planetary-mass companion found in Taurus. The other companion, KOINTREAU-2b (projected separation 560 au), has a spectral t

Sparse Mixtures of Experts (MoEs) are typically trained to operate at a fixed sparsity level, e.g. $k$ in a top-$k$ gating function. This global sparsity level determines an operating point on the accuracy/latency curve; currently, meeting multiple efficiency targets means training and maintaining multiple models. This practice complicates serving, increases training and maintenance costs, and limits flexibility in meeting diverse latency, efficiency, and energy requirements. We show that pretrained MoEs are more robust to runtime sparsity shifts than commonly assumed, and introduce MoE-PHDS ({\bf P}ost {\bf H}oc {\bf D}eclared {\bf S}parsity), a lightweight SFT method that turns a single checkpoint into a global sparsity control surface. PHDS mixes training across sparsity levels and anchors with a short curriculum at high sparsity, requiring no architectural changes. The result is predictable accuracy/latency tradeoffs from one model: practitioners can ``dial $k$'' at inference time without swapping checkpoints, changing architecture, or relying on token-level heuristics. Experiments on OLMoE-1B-7B-0125, Qwen1.5-MoE-A2.7B, and proprietary models fit on multiple operating points s

Wide separation gas giant planets present a challenge to current planet formation theories, and the detection and characterisation of these systems enables us to constrain their formation pathways. The WIde Separation Planets In Time (WISPIT) survey aims to detect and characterise wide separation planetary-mass companions over a range of ages from <5 to 20 Myr around solar-type host stars at distances of 75-500 (median: 140) parsecs. The WISPIT survey carries out two 5 minute H-band exposures with the VLT/SPHERE instrument and IRDIS camera, separated by at least six months to identify co-moving companions via proper motion analysis. These two H-band observations in combination with a follow-up Ks-band observation were used to determine the colour-magnitude of the co-moving companions and to derive their masses by comparing to AMES-COND and AMES-DUSTY evolutionary tracks. We report the discovery of WISPIT 1b and WISPIT 1c, two gas giant exoplanets that are co-moving with the stellar binary WISPIT 1, which itself consists of a K4 star and M5.5 star in a multi-decadal orbit. The planets are at projected separations of 338 au and 840 au and have masses of 10 Mj and 4 Mj respectively

Grain-surface chemistry plays a crucial role in the formation of molecules of astrobiological interest, including H$_{2}$S and complex organic molecules (COMs). They are commonly observed in the gas phase toward star-forming regions, but their detection in ices remains limited. Combining gas-phase observations with chemical modeling is therefore essential for advancing our understanding of their chemistry. In this paper we investigate the factors that promote or hinder molecular complexity combining gas-phase observations of CH$_{3}$OH, H$_{2}$S, OCS, N$_{2}$H$^{+}$, and C$^{18}$O with chemical modeling in two dense cores: Barnard-1b and IC348. We observed millimeter emission lines of CH$_{3}$OH, H$_{2}$S, OCS, N$_{2}$H$^{+}$, and C$^{18}$O along strips using the IRAM 30m and Yebes 40m telescopes. We used the gas-grain chemical model \texttt{Nautilus} to reproduce the observed abundance profiles adjusting parameters such as initial sulfur abundances and binding energies. H$_{2}$S, N$_{2}$H$^{+}$ and C$^{18}$O gas-phase abundances vary up to one order of magnitude towards the extinction peak. CH$_{3}$OH abundance remains quite uniform. These abundances can only be reproduced assumin

We present high-resolution emission spectroscopy observations of the ultra-hot Jupiter MASCARA-1b with CRIRES+ in the K-band, covering the post-eclipse phases of its orbit. These observations complement previously published pre-eclipse data. The stellar and telluric features were removed using SysRem, and the planetary signal was analysed with the cross-correlation technique. After confirming the presence of chemical species in our atmospheric model, we combined the pre- and post-eclipse datasets for a joint analysis. By employing a Bayesian retrieval framework, this joint retrieval enabled us to constrain the spatially varying temperature-pressure (T-P) profile and atmospheric carbon-to-oxygen (C/O) ratio. We detected strong emission signatures of CO and H$_2$O in the post-eclipse and combined datasets. While a well-mixed retrieval model results in a super-solar C/O, allowing for vertically varying chemistry yields C/O values consistent with solar. The retrieved parameters are not only consistent across the datasets but also across different chemical regimes. We did not identify any significant velocity shifts between the detected species or across the datasets, which could otherw

We present EfficientViT-SAM, a new family of accelerated segment anything models. We retain SAM's lightweight prompt encoder and mask decoder while replacing the heavy image encoder with EfficientViT. For the training, we begin with the knowledge distillation from the SAM-ViT-H image encoder to EfficientViT. Subsequently, we conduct end-to-end training on the SA-1B dataset. Benefiting from EfficientViT's efficiency and capacity, EfficientViT-SAM delivers 48.9x measured TensorRT speedup on A100 GPU over SAM-ViT-H without sacrificing performance. Our code and pre-trained models are released at https://github.com/mit-han-lab/efficientvit.

Exoplanet exploration has revealed that many$\unicode{x2013}$perhaps most$\unicode{x2013}$terrestrial exoplanets formed with substantial H$_2$-rich envelopes, seemingly in contrast to solar system terrestrials, for which there is scant evidence of long-lived primary atmospheres. It is not known how a long-lived primary atmosphere might affect the subsequent habitability prospects of terrestrial exoplanets. Here, we present a new, self-consistent evolutionary model of the transition from primary to secondary atmospheres. The model incorporates all Fe-C-O-H-bearing species and simulates magma ocean solidification, radiative-convective climate, thermal escape, and mantle redox evolution. For our illustrative example TRAPPIST-1, our model strongly favors atmosphere retention for the habitable zone planet TRAPPIST-1e. In contrast, the same model predicts a comparatively thin atmosphere for the Venus-analog TRAPPIST-1b, which would be vulnerable to complete erosion via non-thermal escape and is consistent with JWST observations. More broadly, we conclude that the erosion of primary atmospheres typically does not preclude surface habitability, and frequently results in large surface water

We present the first analysis of JWST near-infrared spectroscopy of stellar flares from TRAPPIST-1 during transits of rocky exoplanets. Four flares were observed from 0.6--2.8 $μ$m with NIRISS and 0.6--3.5 $μ$m with NIRSpec during transits of TRAPPIST-1b, f, and g. We discover P$α$ and Br$β$ line emission and characterize flare continuum at wavelengths from 1--3.5 $μ$m for the first time. Observed lines include H$α$, P$α$-P$ε$, Br$β$, He I $λ$0.7062$μ$m, two Ca II infrared triplet (IRT) lines, and the He I IRT. We observe a reversed Paschen decrement from P$α$-P$γ$ alongside changes in the light curve shapes of these lines. The continuum of all four flares is well-described by blackbody emission with an effective temperature below 5300 K, lower than temperatures typically observed at optical wavelengths. The 0.6--1 $μ$m spectra were convolved with the TESS response, enabling us to measure the flare rate of TRAPPIST-1 in the TESS bandpass. We find flares of 10$^{30}$ erg large enough to impact transit spectra occur at a rate of 3.6$\substack{+2.1 \\ -1.3}$ flare d$^{-1}$, $\sim$10$\times$ higher than previous predictions from K2. We measure the amount of flare contamination at 2 $μ$