HiMoE-VLA: Hierarchical Mixture-of-Experts for Generalist Vision-Language-Action Policies

Bei Liu; Haidong Cao; Jiaolong Yang; Xiangyu Zheng; Yaobo Liang; Yichao Shen; Yu-Gang Jiang; Zhiying Du; Zhiyuan Feng; Zuxuan Wu

REVIEW 4 major objections 5 minor 8 cited by

The paper claims that a hierarchical mixture-of-experts action module, with action-space experts at the input/output boundaries and balancing experts at adjacent layers, converts the negative transfer caused by heterogeneous robot data into

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-03 18:18 UTC pith:N7MKYG4N

load-bearing objection Genuinely new two-level MoE action module that moves CALVIN/LIBERO numbers and earns a serious referee, but the central positive-transfer claim is under-specified and the paper has a few fixable reporting inconsistencies. the 4 major comments →

arxiv 2512.05693 v2 pith:N7MKYG4N submitted 2025-12-05 cs.RO cs.AI

HiMoE-VLA: Hierarchical Mixture-of-Experts for Generalist Vision-Language-Action Policies

Zhiying Du , Bei Liu , Yaobo Liang , Yichao Shen , Haidong Cao , Xiangyu Zheng , Zhiyuan Feng , Zuxuan Wu

show 2 more authors

Jiaolong Yang Yu-Gang Jiang

This is my paper

classification cs.RO cs.AI

keywords hierarchical mixture-of-expertsvision-language-action policyheterogeneous robot datanegative transferaction-space specializationflow matchingload balancinggeneralist manipulation

verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Most robot policies trained on large mixed datasets suffer when different robots use different action spaces: joint-angle commands versus end-effector commands, single-arm versus dual-arm. This paper argues that the conflict is architectural, not inherent, and introduces HiMoE-VLA, whose action module places specialized experts at the boundaries to absorb action-space differences, balancing experts next to them to absorb broader embodiment and observation variation, and dense transformer blocks in the middle to build shared representations. Two auxiliary losses—a contrastive action-space regularization and a load-balancing regularization—push the experts toward specialization and balanced use. On simulation benchmarks and real single-arm and dual-arm robots, the model reports the highest success rates among the compared methods; in a controlled from-scratch co-training test, it turns the negative transfer seen in baselines into a positive gain. The paper's conclusion is that heterogeneous robotic data can be unified into one policy if heterogeneity is handled at the right layers.

Core claim

The central claim is that the main obstacle to training one vision-language-action policy on diverse robot data is not the data itself but the way a shared dense action module mixes it. When action spaces differ—joint angles versus end-effector deltas, single-arm versus dual-arm—the authors show that co-training degrades performance, with a baseline dropping from 3.81 to 3.55 on a long-horizon benchmark. Their HiMoE-VLA replaces the dense module with a hierarchical mixture of experts: Action-Space MoE layers at the input and output boundaries specialize computation for each action space; Heterogeneity-Balancing MoE layers in adjacent layers absorb residual variation in embodiments, sensors,

What carries the argument

Hierarchical Mixture-of-Experts (HiMoE) action module: a layered action network in which boundary layers use Action-Space MoE to specialize per action space, adjacent layers use Heterogeneity-Balancing MoE to spread broader variation across experts, and middle dense transformer blocks consolidate shared representations. Routing uses top-4 selection among 32 experts, with a contrastive Action-Space Regularization aligning experts routed to the same action-space token and a load-balancing regularization aligning expected gating scores with realized routing frequencies. Training uses a flow-matching objective, and all states and actions are projected into a fixed 24-dimensional vector with vali

Load-bearing premise

Everything rests on the fixed 24-dimensional unified vector: if a robot's true action space cannot be squeezed into 8 end-effector plus 16 joint-angle numbers without losing information, the experts will route on a distorted input and the transfer gains may vanish.

What would settle it

Co-train HiMoE-VLA on a robot with an action space that exceeds 24 dimensions—for example a 7-DoF arm with a multi-finger hand, where joint commands need more than 16 entries—and compare against a baseline with a dedicated head for that robot. If the unified-vector version does not beat the dedicated head (or at least match the positive-transfer pattern reported for 24-dim spaces), the load-bearing premise is the fixed representation, not the MoE hierarchy.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

If correct, one policy can be pre-trained on a large mixture of single-arm and dual-arm, joint-space and end-effector-space demonstrations without needing dataset-specific heads or manual embodiment labels.
Co-training on heterogeneous action spaces becomes a source of improvement rather than a cause of degradation, at least within the range of the fixed 24-dimensional action/state representation.
The hierarchy gives a concrete recipe: specialize at the boundaries, balance in adjacent layers, share in the middle; ablations show removing either MoE stage or either regularization erodes the gain.
Performance gains on long-horizon tasks and unseen objects suggest the design supports generalization, not just fitting to the training distribution.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

One extension the paper leaves implicit: the same boundary-specialize/middle-share pattern could be applied to other multi-domain sequence models, where the contrastive regularization would act as an automatic classifier of action or control modes.
The fixed 24-dimensional vector suggests a scaling path: as new embodiments are added, the vector must grow and the validity masks must be re-learned; an implicit, testable prediction is that the number of experts should grow with the number of distinct action spaces rather than with dataset size.
The load-balancing objective could be replaced by an adaptive capacity-allocation rule that lets the router discover how many action spaces actually exist, removing the need to pre-specify a fixed top-k.
A direct test of the representation's ceiling would be to feed the same hierarchy with a learned per-embodiment embedding instead of the fixed vector; if that variant matches or exceeds HiMoE, the unified vector, not the MoE structure, is the operative limit.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Desk Editor's Note

Genuinely new two-level MoE action module that moves CALVIN/LIBERO numbers and earns a serious referee, but the central positive-transfer claim is under-specified and the paper has a few fixable reporting inconsistencies.

read the letter

You should know this paper is better than the usual VLA architecture salad. The two-level hierarchy—Action-Space MoE at the boundaries, Heterogeneity-Balancing MoE next to them, dense transformer in the middle—is a real design idea, and the two regularizers (contrastive for action-space specialization, load-balancing for the deeper experts) are not just tacked on; the ablations show each piece earns its keep. The gains on CALVIN (3.97 vs 3.76 for π0) and LIBERO (97.8 vs 94.2 for π0) are solid, and the real-world results, while small, are consistent. Code and models are public, which makes this reproducible. I would not desk-reject this.

The soft spots are real but mostly addressable. The biggest one is Table 7, the direct evidence for the headline claim that HiMoE turns negative transfer into positive transfer. The table compares training on CALVIN-D alone with co-training on ABC+ D, but it only shows success on the D side. We don't see the ABC-side success or an ABC-only baseline. So the +0.186 for the full model shows that adding ABC data helps D—that is positive transfer in one direction—but it does not show the model is a better generalist on both action spaces. The stress-test note is right: the claim as written is one-sided. Also, there are no error bars or significance tests anywhere, and the real-world trials are 15–18 per condition. That is small but not fatal for a robotics paper; still, it needs to be stated.

There are also two numerical inconsistencies that look sloppy rather than malicious: the main text says ALOHA data is 1.6M frames while Appendix A.1 says 36.28M (a 20x difference), and Appendix B says top-k=8 while the main text and Table 9 say top-k=4, with Table 9 showing K=8 as unstable. The uncited "GR00T-Like" baseline should be either cited or renamed. The 24-dimensional fixed action vector with zero-padding is a pragmatic choice, but it does limit the claim of generality to robots that fit in that layout—worth a sentence in the limitations.

Net: the architecture is plausible, the empirical support is decent but not as tight as the abstract implies, and the missing numbers are easy to supply. This deserves a serious referee. I'd send it out with a request for variance reporting, the ABC-side of Table 7, and a cleanup of the inconsistencies.

Would I bring it to reading group? Yes, mainly to argue about whether the MoE hierarchy is the reason for the gains or just extra capacity. And I'd cite it in my own work if I were building on heterogeneous action spaces.

Referee Report

4 major / 5 minor

Summary. The paper proposes HiMoE-VLA, a 4B vision-language-action model whose action module is organized as a hierarchy of mixture-of-experts layers: Action-Space MoE layers at the boundaries specialize computation for different action spaces (e.g., joint-angle vs. end-effector control), Heterogeneity-Balancing MoE layers in adjacent layers handle broader embodiment/sensor/scene variability, and dense Transformer blocks in the middle integrate shared representations. Two auxiliary losses are introduced: a contrastive Action-Space Regularization and a load-balancing Heterogeneity-Balancing Regularization. The model is pretrained on OXE and Aloha data, then fine-tuned and evaluated on CALVIN, LIBERO, and real xArm7 and ALOHA robots; the paper reports state-of-the-art results on these benchmarks and claims that, under heterogeneous co-training, HiMoE turns negative transfer into positive transfer. Code and models are publicly released.

Significance. If the reported results are correct, the paper makes a practically valuable contribution: it offers an architectural treatment of action-space and embodiment heterogeneity in VLAs, with extensive benchmark coverage, public code and models, and ablations of the main design choices. The layered AS-MoE/HB-MoE idea is clean and the routing visualizations are a useful sanity check. However, the paper's strongest conceptual claim, that heterogeneous co-training becomes positively transferable, rests on an under-specified comparison in Table 7, and the empirical sections generally omit uncertainty estimates. As presented, the evidence is suggestive but not yet sufficient for the strong conclusions drawn in the abstract and Section 4.3.

major comments (4)

[§4.3, Table 7] The headline claim that the full HiMoE model 'not only avoids negative transfer but also improves when trained with heterogeneous data' is not established by Table 7 as presented. The columns 'D (Joint)' and 'ABC (EEF) + D (Joint)' do not state the evaluation domain. If the reported sums are CALVIN-D success (the CALVIN metric from Table 1), then the +0.186 gain for Full HiMoE (3.826 → 4.012) only shows that adding ABC data helps performance on D; there is no ABC-only baseline and no co-trained ABC success rate, so the claim of positive transfer on the ABC side is unverified. If the sums are intended as an aggregate over ABC and D, then the D-only baseline is not matched to the two-environment co-trained column. Please report ABC-side success for D-only, ABC-only, and co-trained models, and include multiple seeds.
[§4.2–§4.3, Tables 3–9] No variance, confidence intervals, or significance tests are reported anywhere. Real-world evaluations use 15–18 trials per condition (Appendix A.2), and several headline differences are comparable to plausible binomial noise, e.g., Table 4 overall 63.7% vs. 54.2% for π0, or Table 5 single-arm Sum. 67.6% vs. 55.9% for π0. Simulation results also appear to be single-seed. Adding standard errors, per-seed results, or a significance statement is necessary before 'consistently surpasses' and 'performance boost' can be accepted as quantitative claims.
[§4 Implementation Details vs. Appendix B and Table 9] The final Mixture-of-Experts configuration is ambiguous. Section 4 states N=32 experts with top-K=4 and reports that 'very high routing widths (K=8) cause instability.' Appendix B, however, states 'We set the number of experts to N=32 with a top-k routing of 8,' and refers to a table that shows this configuration outperforms alternatives. Table 9, the actual N/K ablation, contains no K=8 row at all. Since N and K are core architectural hyperparameters and the reported SOTA results depend on this choice, the paper must state the exact configuration used for the main results and make the ablation table consistent.
[§4 Pre-training Dataset; Appendix A.1] The pretraining data sizes are internally inconsistent. Section 4 reports OXE 22.5M frames + Aloha 1.6M frames = 24.1M frames, while Appendix A.1 states that the Aloha datasets contribute 36.28M frames. Table 10 lists the Aloha mixture ratio as 10.4%, which matches neither 1.6/24.1 ≈ 6.6% nor 36.28/(22.5+36.28) ≈ 61.7%. These numbers must be corrected because the pretraining mixture is a central part of the experimental setup and the current text is not reproducible.

minor comments (5)

[Abstract; §4.1; Table 1; Table 6(b)] The CALVIN headline number is reported as 3.98 in the abstract, 3.94 in the text, 3.967 in Table 1, and 4.012 in Table 6(b). Please align these numbers and clarify which evaluation condition each refers to.
[§3.3, Eq. (4)] The AS-Reg objective is under-specified: 'pairs of experts (i,j) assigned to the same action-space token' needs a formal definition of how i and j are selected for each token and how 'assigned' relates to the routing decisions. As written, the positive-pair selection is ambiguous.
[§4.3, Table 6(a) text] The text says that removing pretrained weights (w/o pretrain) leads to a 'more notable drop' than removing MoE re-initialization (w/o init), but the sums are 3.826 and 3.827, respectively. Either report the per-length numbers that support the qualitative claim or rephrase.
[Appendix B] The sentence 'As shown in Table 6, this configuration consistently outperforms...' appears to refer to Table 9, since Table 6 is the ablation of initialization and heterogeneity-handling methods, not the N/K sweep.
[Figure 1] Figure 1 labels the end-effector action space as '6D EEF Poses,' while the text and Appendix A.2 state an 8-dimensional end-effector vector (6D pose plus presumably gripper dimensions). Please make the notation consistent.

Circularity Check

0 steps flagged

No significant circularity: the paper's claims are empirical, benchmarked against external baselines, and no prediction reduces to a fitted value or self-citation by construction.

full rationale

The paper's central claim is empirical: a hierarchical Mixture-of-Experts action module (AS-MoE at boundaries, HB-MoE adjacent, with contrastive and load-balancing regularizations) improves heterogeneous co-training and avoids negative transfer. The auxiliary losses in Eqs. (4)-(7) do enforce expert specialization and balanced routing by construction, but that is their stated training mechanism, not a result being passed off as a prediction. The claimed gains are assessed against external baselines and ablations on CALVIN, LIBERO, xArm7, and ALOHA (Tables 1-9), and the code is public. The fixed 24-dimensional unified action/state vector with zero-padding and validity masks is an input representation, not an output derived from itself. Hyperparameters (N=32, K=4, λ_AS=0.002, λ_HB=0.001) were selected through ablations on CALVIN, which is a form of target-benchmark tuning that can inflate results but is not circular in the paper's equations—no fitted value is renamed as a prediction. The paper's self-citations, such as CogACT (Li et al., 2024), appear as related work or baselines and are not load-bearing for the central architecture. Appendix D's stated limitations concern feature redundancy and model scale, not a conceded circularity. Concerns about Table 7—the under-specified evaluation domain, missing ABC-side success and variance, and possible single-seed comparison—are evidence-quality or correctness issues, not evidence that the derivation reduces to its inputs. No step meets the quoted-reduction standard required for a circularity finding.

Axiom & Free-Parameter Ledger

2 free parameters · 5 axioms · 0 invented entities

The central claim rests on an empirical architecture, not a derivation. The free parameters are hyperparameters selected on the evaluation benchmarks. The axioms are the input representation, the effectiveness of the two regularizers, and the validity of flow-matching and data mixture. No new physical entities are invented; AS-MoE and HB-MoE are architectural modules, not entities with independent falsifiable handles.

free parameters (2)

Number of experts N and top-k K = N=32, K=4 (main text); App. B says K=8
Chosen based on Table 9 ablation on CALVIN; performance varies with these values and the paper reports the best config.
Regularization coefficients lambda_AS and lambda_HB = lambda_AS=0.002, lambda_HB=0.001
Set by hand 'following best practices'; not derived. They trade off specialization and balancing, affecting final CALVIN/LIBERO scores.

axioms (5)

domain assumption A fixed 24-dimensional state-action vector with zero-padding and validity masks is a sufficient encoding for all heterogeneous robots in the training set.
Introduced in Section 3.2.2 and used in Section 4; if the layout loses information for some embodiment, the entire policy is degraded.
domain assumption The contrastive AS-Reg objective with cosine similarity on expert scores induces the intended action-space specialization.
Eq. (4)-(5); the paper does not theoretically prove that the loss leads to the desired routing, and the Single-MoE+Reg ablation suggests the loss can be ineffective in a single MoE.
domain assumption The load-balancing HB-Reg objective (fi*Pi) aligns empirical routing frequencies with expected probabilities in a way that improves abstraction.
Eq. (6)-(7); standard in MoE literature, but the specific behavioral effect on heterogeneity abstraction is only empirically shown.
standard math Flow-matching (Lipman et al., 2022) with Beta-distributed timestep sampling is a valid action-generation objective.
Adopted from prior work (Pi0, Black et al. 2024); not an assumption unique to this paper.
domain assumption The OXE+ALOHA pretraining mixture is representative of general heterogeneous robot data beyond the evaluation benchmarks.
Used in Section 4; the paper does not analyze whether performance depends on the specific data mixture.

pith-pipeline@v1.3.0-alltime-deepseek · 19668 in / 13755 out tokens · 114250 ms · 2026-08-03T18:18:37.197444+00:00 · methodology

0 comments

read the original abstract

Generalist vision--language--action (VLA) policies are typically trained on heterogeneous mixtures of robot demonstrations spanning diverse embodiments, action spaces, and observation configurations. Modeling such heterogeneity with a shared dense action module can induce negative transfer, particularly when action spaces or visual observations differ across data sources. We address this issue with HiMoE-VLA, a VLA framework built around a Hierarchical Mixture-of-Experts (HiMoE) action module. HiMoE uses Action-Space MoE layers at the input/output boundaries to specialize computation for distinct action spaces, Heterogeneity-Balancing MoE layers in neighboring layers to provide balanced capacity for residual variation in observations, scenes, and embodiments, and dense Transformer blocks in the middle to integrate shared representations. Two auxiliary objectives further guide this hierarchy: a contrastive Action-Space Regularization objective for boundary specialization and a load-balancing objective for stable expert utilization. HiMoE-VLA reaches 3.98 on CALVIN, 98.0\% on LIBERO, and 75.0\% and 63.7\% average success on real xArm7 and ALOHA tasks; under controlled heterogeneous co-training, it turns the negative transfer observed in strong baselines into positive transfer. The code and models are publicly available at https://github.com/ZhiyingDu/HiMoE-VLA.

Figures

Figures reproduced from arXiv: 2512.05693 by Bei Liu, Haidong Cao, Jiaolong Yang, Xiangyu Zheng, Yaobo Liang, Yichao Shen, Yu-Gang Jiang, Zhiying Du, Zhiyuan Feng, Zuxuan Wu.

**Figure 2.** Figure 2: Detailed structure of the Hierarchical Mixture-of-Experts (HiMoE). The architecture fol [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** Figure 3: Qualitative examples of real-world executions on (left) the single-arm xArm7 and (right) [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗

**Figure 4.** Figure 4: Expert Activation Heatmap of AS-MoE. Expert 0 Expert 1 Expert 2 Expert 3 Expert 4 Expert 5 Expert 6 Expert 7 CALVIN JOINT CALVIN EEF LIBERO EEF 201 391 499 297 227 240 170 175 310 367 422 275 208 190 271 157 144 144 185 184 386 495 385 337 Expert Activation Heatmap of HB-MoE 150 200 250 300 350 400 450 [PITH_FULL_IMAGE:figures/full_fig_p019_4.png] view at source ↗

**Figure 5.** Figure 5: Expert Activation Heatmap of HB-MoE. used to train general VLMs. Increasing the model capacity, incorporating stronger pretrained VLMs, and training on more extensive and varied robotics datasets could further enhance generalization and robustness. Investigating how model scale interacts with embodiment-specific variability in learning will be an important topic for future research. E MORE VISUALIZATIONS T… view at source ↗

**Figure 6.** Figure 6: Overview of tasks on singe-arm robot Xarm, including seen tasks and unseen tasks (unseen [PITH_FULL_IMAGE:figures/full_fig_p020_6.png] view at source ↗

**Figure 7.** Figure 7: Overview of tasks on dual-arm robot ALOHA, including seen tasks and unseen tasks [PITH_FULL_IMAGE:figures/full_fig_p020_7.png] view at source ↗

discussion (0)

Forward citations

Cited by 8 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

Route by Kinematics, Act by Observation: Kinematics-Supervised Expert Routing in MoE-Augmented VLA
cs.RO 2026-07 conditional novelty 6.0

Supervising MoE routers with offline kinematic trajectory clusters, then routing from vision-language alone, lifts VLA manipulation success by roughly 20%+ over dense and implicit-MoE baselines.
PAMAE: Phase-Aware-MoE Action Experts Towards Reliable Flow-Matching Vision-Language-Action Policies
cs.RO 2026-06 unverdicted novelty 6.0

PAMAE adds a phase-aware router and expert mixture to flow-matching VLA models, yielding up to 9.2% higher task success on multi-stage manipulation simulations via two-stage training.
UniFS: Unified Fast-to-Slow Hierarchical Architecture for Vision-Language-Action Models
cs.RO 2026-06 unverdicted novelty 6.0

UniFS achieves 98.3% success on LIBERO with 2.1x lower latency than prior fast-slow VLA models by stratifying VLM layer update frequencies, inverting latent interactions, and applying multi-level supervision.
SARM2: Multi-Task Stage Aware Reward Modeling for Self Improving Robotic Manipulation
cs.RO 2026-06 unverdicted novelty 6.0

SARM2 presents RM, a multi-task stage-aware reward model achieving 80% lower value-estimation MSE, which when used in SPIRAL boosts manipulation task success from ~50% to near-perfect on several benchmarks.
TORL-VLA: Tactile Guided Online Reinforcement Learning for Contact-Rich Manipulation
cs.RO 2026-06 unverdicted novelty 6.0

TORL-VLA couples a tactile wrench-aware VLA policy with a lightweight online RL module and an intervention-censored critic to improve success and efficiency on contact-rich robotic tasks.
HEX: Humanoid-Aligned Experts for Cross-Embodiment Whole-Body Manipulation
cs.RO 2026-04 unverdicted novelty 6.0

HEX is a new framework with humanoid-aligned state representation, mixture-of-experts proprioceptive predictor, history tokens, and residual-gated fusion that achieves state-of-the-art success and generalization on re...
HEX: Humanoid-Aligned Experts for Cross-Embodiment Whole-Body Manipulation
cs.RO 2026-04 unverdicted novelty 6.0

HEX introduces a state-centric framework with humanoid-aligned representations and mixture-of-experts proprioceptive prediction for coordinated whole-body control on bipedal humanoids.
From Foundation to Application: Improving VLA Models in Practice
cs.RO 2026-07 conditional novelty 4.0

LingBot-VLA 2.0 combines 60k hours of multi-embodiment pretraining data, an expanded whole-body action space, and dual-query distillation from depth and video teachers to improve VLA performance on GM-100 and long-hor...

Reference graph

Works this paper leans on

4 extracted references · cited by 7 Pith papers

[1]

write newline

" write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...
[2]

@esa (Ref

\@ifxundefined[1] #1\@undefined \@firstoftwo \@secondoftwo \@ifnum[1] #1 \@firstoftwo \@secondoftwo \@ifx[1] #1 \@firstoftwo \@secondoftwo [2] @ #1 \@temptokena #2 #1 @ \@temptokena \@ifclassloaded agu2001 natbib The agu2001 class already includes natbib coding, so you should not add it explicitly Type <Return> for now, but then later remove the command n...
[3]

\@lbibitem[] @bibitem@first@sw\@secondoftwo \@lbibitem[#1]#2 \@extra@b@citeb \@ifundefined br@#2\@extra@b@citeb \@namedef br@#2 \@nameuse br@#2\@extra@b@citeb \@ifundefined b@#2\@extra@b@citeb @num @parse #2 @tmp #1 NAT@b@open@#2 NAT@b@shut@#2 \@ifnum @merge>\@ne @bibitem@first@sw \@firstoftwo \@ifundefined NAT@b*@#2 \@firstoftwo @num @NAT@ctr \@secondoft...
[4]

@open @close @open @close and [1] URL: #1 \@ifundefined chapter * \@mkboth \@ifxundefined @sectionbib * \@mkboth * \@mkboth\@gobbletwo \@ifclassloaded amsart * \@ifclassloaded amsbook * \@ifxundefined @heading @heading NAT@ctr thebibliography [1] @ \@biblabel @NAT@ctr \@bibsetup #1 @NAT@ctr @ @openbib .11em \@plus.33em \@minus.07em 4000 4000 `\.\@m @bibit...

[1] [1]

write newline

" write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...

[2] [2]

@esa (Ref

\@ifxundefined[1] #1\@undefined \@firstoftwo \@secondoftwo \@ifnum[1] #1 \@firstoftwo \@secondoftwo \@ifx[1] #1 \@firstoftwo \@secondoftwo [2] @ #1 \@temptokena #2 #1 @ \@temptokena \@ifclassloaded agu2001 natbib The agu2001 class already includes natbib coding, so you should not add it explicitly Type <Return> for now, but then later remove the command n...

[3] [3]

\@lbibitem[] @bibitem@first@sw\@secondoftwo \@lbibitem[#1]#2 \@extra@b@citeb \@ifundefined br@#2\@extra@b@citeb \@namedef br@#2 \@nameuse br@#2\@extra@b@citeb \@ifundefined b@#2\@extra@b@citeb @num @parse #2 @tmp #1 NAT@b@open@#2 NAT@b@shut@#2 \@ifnum @merge>\@ne @bibitem@first@sw \@firstoftwo \@ifundefined NAT@b*@#2 \@firstoftwo @num @NAT@ctr \@secondoft...

[4] [4]

@open @close @open @close and [1] URL: #1 \@ifundefined chapter * \@mkboth \@ifxundefined @sectionbib * \@mkboth * \@mkboth\@gobbletwo \@ifclassloaded amsart * \@ifclassloaded amsbook * \@ifxundefined @heading @heading NAT@ctr thebibliography [1] @ \@biblabel @NAT@ctr \@bibsetup #1 @NAT@ctr @ @openbib .11em \@plus.33em \@minus.07em 4000 4000 `\.\@m @bibit...