Pith Number

pith:C5W5P6KV

pith:2021:C5W5P6KVGEHNQOJOI7SI2L27Z2

not attested not anchored not stored refs resolved

DeBERTaV3: Improving DeBERTa using ELECTRA-Style Pre-Training with Gradient-Disentangled Embedding Sharing

Jianfeng Gao, Pengcheng He, Weizhu Chen

DeBERTaV3 replaces masked language modeling with replaced token detection and introduces gradient-disentangled embedding sharing to raise accuracy on natural language understanding benchmarks.

arxiv:2111.09543 v4 · 2021-11-18 · cs.CL · cs.LG

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

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{C5W5P6KVGEHNQOJOI7SI2L27Z2}

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

the DeBERTaV3 Large model achieves a 91.37% average score, which is 1.37% over DeBERTa and 1.91% over ELECTRA, setting a new state-of-the-art (SOTA) among the models with a similar structure.

C2weakest assumption

That the observed gains come from the gradient-disentangled sharing rather than from other unstated differences in training schedule, data order, or hyper-parameters between the new runs and the cited DeBERTa/ELECTRA baselines.

C3one line summary

DeBERTaV3 improves DeBERTa by switching to replaced token detection pre-training and using gradient-disentangled embedding sharing, reaching 91.37% on GLUE and new SOTA on XNLI zero-shot.

References

29 extracted · 29 resolved · 8 Pith anchors

[1] Language Models are Few-Shot Learners 2005 · arXiv:2005.14165

[2] Semeval-2017 task 1: Semantic textual similarity-multilingual and cross-lingual focused evaluation 2017 · arXiv:1708.00055

[3] Xlm-e: Cross-lingual language model pre-training via electra

[4] Xnli: Evaluating cross-lingual sentence representations 2018

[5] Bert: Pre-training of deep bidirectional transformers for language understanding 2019

Formal links

2 machine-checked theorem links

Cited by

41 papers in Pith

Talk Less, Fly Lighter: Autonomous Semantic Compression for UAV Swarm Communication via LLMs

Survey in Characterizing Semantic Change

GigaCheck: Detecting LLM-generated Content via Object-Centric Span Localization

Semantic Embeddings of Chemical Elements for Enhanced Materials Inference and Discovery

Fact-Checking with Contextual Narratives: Leveraging Retrieval-Augmented LLMs for Social Media Analysis

Receipt and verification

First computed	2026-05-17T23:38:52.519089Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

176dd7f955310ed8392e47e48d2f5fceb0214103eb77daf9ae84779d57b047bb

Aliases

arxiv: 2111.09543 · arxiv_version: 2111.09543v4 · doi: 10.48550/arxiv.2111.09543 · pith_short_12: C5W5P6KVGEHN · pith_short_16: C5W5P6KVGEHNQOJO · pith_short_8: C5W5P6KV

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/C5W5P6KVGEHNQOJOI7SI2L27Z2 \
  | 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: 176dd7f955310ed8392e47e48d2f5fceb0214103eb77daf9ae84779d57b047bb

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "f35e93f6ffbc86590f62ab257c29e8323fbfd90db0896f8a2e22f75307f307a9",
    "cross_cats_sorted": [
      "cs.LG"
    ],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "primary_cat": "cs.CL",
    "submitted_at": "2021-11-18T06:48:00Z",
    "title_canon_sha256": "bbfea01373de9b0c2623a49541e84f761d6cab8f9bc1eb766acce36e5188d917"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2111.09543",
    "kind": "arxiv",
    "version": 4
  }
}