Pith Number

pith:XZ6KU673

pith:2024:XZ6KU673RXJCDFKJNQGACJNLRV

not attested not anchored not stored refs resolved

Auxiliary-Loss-Free Load Balancing Strategy for Mixture-of-Experts

Chenggang Zhao, Damai Dai, Huazuo Gao, Lean Wang, Xu Sun

Mixture-of-Experts models reach higher performance with load balancing that avoids auxiliary loss gradients.

arxiv:2408.15664 v1 · 2024-08-28 · cs.LG · cs.CL

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{XZ6KU673RXJCDFKJNQGACJNLRV}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

Loss-Free Balancing achieves both better performance and better load balance compared with traditional auxiliary-loss-controlled load balancing strategies.

C2weakest assumption

That dynamically updating per-expert biases from recent load statistics will maintain balance across training without introducing instability or unintended routing dynamics.

C3one line summary

Loss-Free Balancing keeps expert loads balanced in MoE models by dynamically adjusting routing-score biases based on recent usage, avoiding auxiliary-loss interference and yielding better performance.

References

32 extracted · 32 resolved · 6 Pith anchors

[1] DeepSeekMoE: Towards Ultimate Expert Specialization in Mixture-of-Experts Language Models 2024 · arXiv:2401.06066

[2] DeepSeek-Coder-V2: Breaking the Barrier of Closed-Source Models in Code Intelligence 2024 · arXiv:2406.11931

[3] William Fedus, Barret Zoph, and Noam M. Shazeer. Switch transformers: Scaling to trillion parameter models with simple and efficient sparsity. J. Mach. Learn. Res., 23: 0 120:1--120:39, 2021. URL http 2021

[4] GShard: Scaling Giant Models with Conditional Computation and Automatic Sharding 2006 · arXiv:2006.16668

[5] Sgdr: Stochastic gradient descent with warm restarts 2016

Formal links

2 machine-checked theorem links

Cited by

39 papers in Pith

FLEX-MoE: Federated Mixture-of-Experts with Load-balanced Expert Assignment for Edge Computing

S$^2$GR: Stepwise Semantic-Guided Reasoning in Latent Space for Generative Recommendation

CP-MoE: Consistency-Preserving Mixture-of-Experts for Continual Learning

DECO: Sparse Mixture-of-Experts with Dense-Comparable Performance on End-Side Devices

Dynamic Chunking for Diffusion Language Models

Receipt and verification

First computed	2026-05-17T23:38:52.511754Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

be7caa7bfb8dd22195496c0c0125ab8d718011f0f205abb2232fbcaf12a94cf6

Aliases

arxiv: 2408.15664 · arxiv_version: 2408.15664v1 · doi: 10.48550/arxiv.2408.15664 · pith_short_12: XZ6KU673RXJC · pith_short_16: XZ6KU673RXJCDFKJ · pith_short_8: XZ6KU673

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/XZ6KU673RXJCDFKJNQGACJNLRV \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: be7caa7bfb8dd22195496c0c0125ab8d718011f0f205abb2232fbcaf12a94cf6

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "de6550001f65ea19f000f8d635df0f482bea715eade709c791379eef9256cdc5",
    "cross_cats_sorted": [
      "cs.CL"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.LG",
    "submitted_at": "2024-08-28T09:31:09Z",
    "title_canon_sha256": "2ba3f20877dadd522332bfbf50455cc66980575dd5bbc31e024cf536d29778ff"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2408.15664",
    "kind": "arxiv",
    "version": 1
  }
}