Pith Number

pith:OJEIIRC6

pith:2023:OJEIIRC6PPGPHOZ4JDA3H4YF3B

not attested not anchored not stored refs resolved

PaLM 2 Technical Report

Aakanksha Chowdhery, Abe Ittycheriah, Alexandre Passos, Alex Castro Ros, Alex Polozov, Alicia Parrish, Ambrose Slone, Andrea Hu, Andrew M. Dai, Andrew Nystrom, Aroma Mahendru, Aurko Roy, Benjamin Lee, Brennan Saeta, Bryan Richter, Ce Zheng, Chang Lan, Christopher A. Choquette-Choo, Cl\'ement Crepy, Colin Cherry, Daniel Smilkov, Daniel Sohn, Dasha Valter, David R. So, Denny Zhou, Dmitry Lepikhin, Emanuel Taropa, Emily Reif, Erica Moreira, Eric Chu, Eric Li, Eric Ni, Ethan Dyer, Fangxiaoyu Feng, Frederick Liu, Gaurav Mishra, Gustavo Hernandez Abrego, Guy Gur-Ari, Hadi Hashemi, Hanzhao Lin, Hyeontaek Lim, Jacob Austin, Jacob Devlin, James Bradbury, Jan Botha, Jeffrey Hui, Jeremy Hurwitz, Jiahui Yu, Jian Li, John Nham, John Wieting, Jonathan H. Clark, Joshua Howland, Joshua Maynez, Junwhan Ahn, Katherine Lee, Kathleen Kenealy, Kathy Meier-Hellstern, Kefan Xiao, Kelvin Xu, Kevin Brooks, Kevin Robinson, Kiran Vodrahalli, Laurent El Shafey, Le Hou, Linting Xue, Lucas Gonzalez, Marcello Maggioni, Marie Pellat, Mark D\'iaz, Mark Omernick, Markus Freitag, Martin Polacek, Matthew Jagielski, Maxim Krikun, Maysam Moussalem, Melvin Johnson, Michael Isard, Michele Catasta, Mostafa Dehghani, Music Li, Nan Du, Orhan Firat, Paige Bailey, Parker Riley, Paul Barham, Pengcheng Yin, Pidong Wang, Qiao Zhang, Rajkumar Samuel, Reiner Pope, Renee Shelby, Rohan Anil, Sebastian Gehrmann, Sebastian Ruder, Shachi Dave, Siamak Shakeri, Siddhartha Brahma, Simon Tokumine, Siyuan Qiao, Slav Petrov, Sneha Kudugunta, Steven Hand, Steven Zheng, Sunipa Dev, Tao Wang, Vedant Misra, Vijay Vasudevan, Vlad Feinberg, Vlad Fienber, Weikang Zhou, Wei Li, Wenhao Jia, Xavier Garcia, Xuezhi Wang, Yaguang Li, Yanping Huang, Yi Tay, Yong Cheng, Yonghui Wu, Yuanzhong Xu, Yuhuai Wu, Yujing Zhang, Yunhan Xu, Zachary Nado, Zhifeng Chen, Zhongtao Liu, Zirui Wang

PaLM 2 raises quality on English, multilingual, and reasoning tasks while cutting inference time and compute compared to PaLM.

arxiv:2305.10403 v3 · 2023-05-17 · cs.CL · cs.AI

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

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{OJEIIRC6PPGPHOZ4JDA3H4YF3B}

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

PaLM 2 has significantly improved quality on downstream tasks across different model sizes, while simultaneously exhibiting faster and more efficient inference compared to PaLM.

C2weakest assumption

That the suite of English, multilingual, and reasoning benchmarks plus the responsible-AI evaluations accurately capture real-world performance and generalization without undisclosed data selection or post-training adjustments.

C3one line summary

PaLM 2 reports state-of-the-art results on language, reasoning, and multilingual tasks with improved efficiency over PaLM.

References

286 extracted · 286 resolved · 32 Pith anchors

[1] Persistent anti-muslim bias in large language models 2021

[2] Akhbardeh, F., Arkhangorodsky, A., Biesialska, M., Bojar, O., Chatterjee, R., Chaudhary, V., Costa-jussa, M. R., Espa \ n a-Bonet, C., Fan, A., Federmann, C., Freitag, M., Graham, Y., Grundkiewicz, R. 2021

[3] Guide to fair pay 2023

[5] Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback 2022 · arXiv:2204.05862

[6] Building machine translation systems for the next thousand languages 2022

Formal links

2 machine-checked theorem links

Cited by

111 papers in Pith

Beware What You Autocomplete: Forensic Attribution of Backdoored Code Completions

Detect, Unlearn, Restore: Defending Text Summarization Models Against Data Poisoning

Evaluating Japanese Dialect Robustness Across Speech and Text-based Large Language Models

Data-driven Machine Learning Cannot Reach Symbolic-level Logical Reasoning -- The Limit of the Scaling Law

SingGuard: A Policy-Adaptive Multimodal LLM Guardrail with Dynamic Reasoning

Receipt and verification

First computed	2026-07-05T06:50:29.521454Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

724884445e7bccf3bb3c48c1b3f305d846cca4ee16af50f662152f0aa95ae36c

Aliases

arxiv: 2305.10403 · arxiv_version: 2305.10403v3 · doi: 10.48550/arxiv.2305.10403 · pith_short_12: OJEIIRC6PPGP · pith_short_16: OJEIIRC6PPGPHOZ4 · pith_short_8: OJEIIRC6

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/OJEIIRC6PPGPHOZ4JDA3H4YF3B \
  | 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: 724884445e7bccf3bb3c48c1b3f305d846cca4ee16af50f662152f0aa95ae36c

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "3a1c4f2d7b48b4ab4b1e818d837be35a2026c4120ade3fe45234bd9c33f7d6d5",
    "cross_cats_sorted": [
      "cs.AI"
    ],
    "license": "http://creativecommons.org/licenses/by-sa/4.0/",
    "primary_cat": "cs.CL",
    "submitted_at": "2023-05-17T17:46:53Z",
    "title_canon_sha256": "ac37805c5ff5e690c3c8d110c47289e3ea3088176ac37d5f2c326d2978887f55"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2305.10403",
    "kind": "arxiv",
    "version": 3
  }
}