LIMA

Closed weights Meta AI,Carnegie Mellon University (CMU),University of Southern California,Tel Aviv University 65B parameters May 2023

No estimate

No hardware requirements for this model

The weights for this model have not been published, so it cannot be downloaded or run on your own hardware at any size. It is reachable only through its provider, and no graphics card changes that.

On record

Full specification

Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.

Origin

Who built this model, where, and when it was published.

Organisation
Meta AI,Carnegie Mellon University (CMU),University of Southern California,Tel Aviv University
Organisation type
Industry,Academia,Academia,Academia
Country
United States of America, Israel
Published
18 May 2023
Authors
Chunting Zhou, Pengfei Liu, Puxin Xu, Srini Iyer, Jiao Sun, Yuning Mao, Xuezhe Ma, Avia Efrat, Ping Yu, Lili Yu, Susan Zhang, Gargi Ghosh, Mike Lewis, Luke Zettlemoyer, Omer Levy

What it does

The problem areas the model was built for. A model can carry several of each.

Domain
Language
Task
Chat, Language modeling/generation
Base model
LLaMA-65B

Size

How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.

Parameters
65B

"We train LIMA (Less Is More for Alignment) using the following protocol. Starting from LLaMa 65B [Touvron et al., 2023], we fine-tune on our 1,000-example alignment training set," according to page 4 of https://arxiv.org/pdf/2305.11206.

Training data
685,300 tokens

The total amount of training data is roughly 750,000 tokens, split over exactly 1,000 sequences.

Epochs
15

Training compute

The arithmetic performed to train the model, measured in floating-point operations. It is a measure of what the training run cost, not of how fast the finished model answers you.

Training compute
5.5 × 10²³ FLOP

Finetune: 4.39e18 FLOP Base model: 5.5e+23 FLOP Total: 4.39e18+5.5e+23=5.5e+23 FLOP

How it was established
Operation counting
Fine-tuning compute
4.4 × 10²⁰ FLOP

“The total amount of training data is roughly 750,000 tokens, split over exactly 1,000 sequences,” according to page 2 of https://arxiv.org/pdf/2305.11206. “We train LIMA (Less Is More for Alignment) using the following protocol. Starting from LLaMa 65B [Touvron et al., 2023], we fine-tune on our 1,000-example alignment training set. To differentiate between each speaker (user and assistant), we introduce a special end-of-turn token (EOT) at the end of each utterance; this token plays the same …

Availability

Whether you can obtain the model and run it on your own hardware, which is what decides if any of the graphics-card figures on this page apply.

Weights
Closed — provider access only
Model access
Unreleased
Training code
Unreleased

How it is classified

Labels the source dataset applies when tracking notable models, and how confident it is in the entry.

Likely above 10²³ FLOP
Yes
Record confidence
Confident
Citations
1,274

Sources

Where this record came from and when it was last checked.

Reference
LIMA: Less Is More for Alignment
Last updated
25 May 2026

What the numbers mean

What this model is

LIMA was published by Meta AI,Carnegie Mellon University (CMU),University of Southern California,Tel Aviv University, in United States of America, in May 2023. industry,Academia,Academia,Academia is the category the publisher falls under.

It works in Language, and is recorded as doing chat, Language modeling/generation.

Its starting point was LLaMA-65B — most models at this scale are adapted from an existing base rather than built from nothing.

Because the weights are not available, none of the hardware figures elsewhere on this site apply to it.

What went into building it

Producing it required around 5.5 × 10²³ FLOP of arithmetic, which is a statement about the training budget rather than about inference.

The training set ran to roughly 685,300 tokens.

Answers

LIMA — common questions

01

What is LIMA used for?

LIMA works in Language, and is recorded as handling chat, Language modeling/generation. These are the areas it was designed around; they describe intent rather than a hard boundary.

02

How much compute was used to train LIMA?

Around 5.5 × 10²³ FLOP. That measures what producing the model cost and says nothing about how quickly it answers once trained — inference speed comes from memory bandwidth, not from the training budget.

03

What GPU do I need to run LIMA?

None. LIMA is a closed model — its weights were never published, so it cannot be downloaded or run on your own hardware at any price. It is reachable only through its provider.

04

Is LIMA open source?

No. LIMA has not had its weights published, so it exists only as a service controlled by its owner.

05

How many parameters does LIMA have?

LIMA has 65B parameters. "We train LIMA (Less Is More for Alignment) using the following protocol. Starting from LLaMa 65B [Touvron et al., 2023], we fine-tune on our 1,000-example alignment training set," according to page 4 of https://arxiv.org/pdf/2305.11206. That figure is the total, and it is what decides how much memory the model needs — roughly half a gigabyte per billion at the compression most people use.

06

Who created LIMA?

LIMA was published by Meta AI,Carnegie Mellon University (CMU),University of Southern California,Tel Aviv University, based in United States of America, categorised as industry,Academia,Academia,Academia.

07

When was LIMA released?

LIMA was published in May 2023. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

Source

Original publication

Record last updated 25 May 2026

The other direction

Looking at it from the other side?

This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.