Pith Number

pith:C4EW5CF5

pith:2024:C4EW5CF5ETZB7Y2CQ4JOCWDXXG

not attested not anchored not stored refs resolved

GaLore: Memory-Efficient LLM Training by Gradient Low-Rank Projection

Anima Anandkumar, Beidi Chen, Jiawei Zhao, Yuandong Tian, Zhangyang Wang, Zhenyu Zhang

GaLore projects full gradients onto low-rank subspaces periodically, cutting optimizer memory by 65.5% while training every parameter of large language models.

arxiv:2403.03507 v2 · 2024-03-06 · 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{C4EW5CF5ETZB7Y2CQ4JOCWDXXG}

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

We demonstrate, for the first time, the feasibility of pre-training a 7B model on consumer GPUs with 24GB memory (e.g., NVIDIA RTX 4090) without model parallel, checkpointing, or offloading strategies.

C2weakest assumption

That periodically recomputed low-rank bases for gradient projection preserve optimization dynamics close enough to full-rank gradients that final model quality remains comparable across pre-training and fine-tuning regimes.

C3one line summary

GaLore performs full-parameter LLM training with up to 65.5% less optimizer memory by projecting gradients onto a low-rank subspace at each step, matching full-rank performance on LLaMA pre-training and RoBERTa fine-tuning.

References

55 extracted · 55 resolved · 9 Pith anchors

[1] Memory efficient adaptive optimization 2019

[2] Belle: Be everyone's large language model engine 2023

[3] Continual learning in low-rank orthogonal subspaces 2020

[4] Non- Convex Projected Gradient Descent for Generalized Low-Rank Tensor Regression 2019

[5] Training Deep Nets with Sublinear Memory Cost 2016 · arXiv:1604.06174

Formal links

2 machine-checked theorem links

Cited by

29 papers in Pith

HTMuon: Improving Muon via Heavy-Tailed Spectral Correction

Proof-of-Learning with Incentive Security

GWT: Scalable Optimizer State Compression for Large Language Model Training

Small Language Models (SLMs) Can Still Pack a Punch: A survey (updated 2026)

Memory-Efficient Differentially Private Training with Gradient Random Projection

Receipt and verification

First computed	2026-05-17T23:38:46.186714Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

17096e88bd24f21fe3428712e15877b9989ea67ba7b88a5a35cb99e74b6d2990

Aliases

arxiv: 2403.03507 · arxiv_version: 2403.03507v2 · doi: 10.48550/arxiv.2403.03507 · pith_short_12: C4EW5CF5ETZB · pith_short_16: C4EW5CF5ETZB7Y2C · pith_short_8: C4EW5CF5

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/C4EW5CF5ETZB7Y2CQ4JOCWDXXG \
  | 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: 17096e88bd24f21fe3428712e15877b9989ea67ba7b88a5a35cb99e74b6d2990

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "7b1eac0584b3aab5267eb2f6d6033547186f3b3a34a6f003bdeae70ede5b6275",
    "cross_cats_sorted": [],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "primary_cat": "cs.LG",
    "submitted_at": "2024-03-06T07:29:57Z",
    "title_canon_sha256": "b3d8e896d144add070f5ec4b0c898932d6e3a058b1cd57804eba277c6a3cfaa2"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2403.03507",
    "kind": "arxiv",
    "version": 2
  }
}