What are the hardware requirements for Sheared LLaMA 2.7B?

Sheared LLaMA 2.7B requires approximately 10.8 GB of VRAM and supports a context window of 4K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Sheared LLaMA 2.7B and how large is it?

Sheared LLaMA 2.7B is developed by princeton-nlp, a model with 2.7b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does Sheared LLaMA 2.7B perform on standard benchmarks?

Sheared LLaMA 2.7B has the following published scores: MMLU 26.92. Compare against reference models on this page or on the LLM Explorer leaderboards.

Sheared LLaMA 2.7B by princeton-nlp — VRAM 10.8GB, 4K context

Name: Sheared LLaMA 2.7B
Author: princeton-nlp

Sheared LLaMA 2.7B is an open-source language model by princeton-nlp. Features: 2.7b LLM, VRAM: 10.8GB, Context: 4K, License: apache-2.0, LLM Explorer Score: 0.15, Arc: 41.7, HellaSwag: 71, MMLU: 26.9, GSM8K: 1.1.

Arxiv:2310.06694 Endpoints compatible Llama Pytorch Region:us Sharded

Model Card on HF 🤗: https://huggingface.co/princeton-nlp/Sheared-LLaMA-2.7B

Sheared LLaMA 2.7B Parameters and Internals

Model Type

language model

Use Cases

Areas:

research, commercial applications

Applications:

reasoning, reading comprehension, language modeling, knowledge intensive tasks

Primary Use Cases:

Various NLP tasks and evaluations based on extensive downstream task metrics.

Additional Notes

Dynamically loads data from different domains during pre-training. Pruned and further pre-trained from meta-llama/Llama-2-7b-hf.

Training Details

Data Sources:

RedPajama dataset

Data Volume:

50 billion tokens

Methodology:

Structured pruning and continued pre-training

LLM Name	Sheared LLaMA 2.7B
Repository 🤗	https://huggingface.co/princeton-nlp/Sheared-LLaMA-2.7B
Model Size	2.7b
Required VRAM	10.8 GB
Updated	2026-08-06
Maintainer	princeton-nlp
Model Type	llama
Model Files	9.9 GB: 1-of-2 0.9 GB: 2-of-2
Model Architecture	LlamaForCausalLM
License	apache-2.0
Context Length	4096
Model Max Length	4096
Transformers Version	4.28.1
Tokenizer Class	LlamaTokenizer
Beginning of Sentence Token	<s>
End of Sentence Token	</s>
Unk Token	<unk>
Vocabulary Size	32000
Torch Data Type	float32

Rank the Sheared LLaMA 2.7B capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to Sheared LLaMA 2.7B

Best Alternatives	Context / RAM	Downloads	Likes
Dopeyshearedplats 2.7B V1	4K / 5.7 GB	214	2
...redplats 2.7B V2 Instruct V0.1	4K / 5.4 GB	5	3
Shearedplats 2.7B V2	4K / 5.7 GB	226	2
Llama Final	4K / 5.5 GB	5	0
DPO VinaLlama Completed	4K / 5.6 GB	8	0
Urllm Ko En 2.7B	4K / 11.2 GB	5	0
...heared Llama 2.7B Deita 6K Sft	4K / 5.4 GB	4	1
Sheared LLaMA 2.7B Weblinx	4K / 5.4 GB	23	2
Sheared LLaMA 2.7B Pruned	4K / 5.4 GB	21	3
Vinallama 2.7B Chat	4K / 5.5 GB	370	16