What are the hardware requirements for Llama 2 13B GPTQ?

Llama 2 13B GPTQ requires approximately 7.3 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 Llama 2 13B GPTQ and how large is it?

Llama 2 13B GPTQ is developed by TheBloke, a model with 13b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does Llama 2 13B GPTQ perform on standard benchmarks?

Llama 2 13B GPTQ has the following published scores: MMLU 54.45. Compare against reference models on this page or on the LLM Explorer leaderboards.

Llama 2 13B GPTQ by TheBloke — VRAM 7.3GB, 4K context

Name: Llama 2 13B GPTQ
Author: TheBloke

Llama 2 13B GPTQ is an open-source language model by TheBloke. Features: 13b LLM, VRAM: 7.3GB, Context: 4K, License: llama2, Quantized, LLM Explorer Score: 0.11, Arc: 59.1, HellaSwag: 81.5, MMLU: 54.5, GSM8K: 11.3.

Arxiv:2307.09288 4-bit Base model:meta-llama/llama-2-... Base model:quantized:meta-llam... Deploy:azure En Facebook Gptq Llama Llama2 Meta Pytorch Quantized Region:us Safetensors

Model Card on HF 🤗: https://huggingface.co/TheBloke/Llama-2-13B-GPTQ

Llama 2 13B GPTQ Parameters and Internals

Model Type

llama

Use Cases

Areas:

commercial, research

Applications:

assistant-like chat, natural language generation

Primary Use Cases:

Chat models for dialogue

Limitations:

English-focused, not tested in all languages., Potentially unpredictable outputs.

Considerations:

Follow specific input formatting to align with intended use cases.

Additional Notes

Compatible with AutoGPTQ and major GPTQ clients. Choose quantization parameters based on hardware needs.

Supported Languages

en (fluent)

Training Details

Data Sources:

Publicly available online data

Data Volume:

2 trillion tokens

Methodology:

Pretrained and fine-tuned using supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF)

Context Length:

4096

Training Time:

January 2023 to July 2023

Hardware Used:

A100-80GB GPUs

Model Architecture:

Optimized transformer architecture.

Safety Evaluation

Methodologies:

Internal evaluations library

Findings:

May produce inaccurate, biased or objectionable responses; testing primarily in English.

Risk Categories:

TruthfulQA, Toxigen

Ethical Considerations:

Before deploying applications, perform safety testing tailored to your use case.

Responsible Ai Considerations

Fairness:

Testing primarily in English, does not guarantee unbiased outputs in all languages.

Transparency:

Evaluation data and results are disclosed.

Accountability:

Developers responsible for application-specific safety testing.

Mitigation Strategies:

Community feedback and iterative improvements.

Input Output

Input Format:

Text prompt

Accepted Modalities:

text

Output Format:

Generated text

Performance Tips:

Select appropriate quantization parameters for VRAM efficiency and accuracy.

Release Notes

Version:

Llama 2 13B

Date:

July 2023

Notes:

Pretrained on 2 trillion tokens with fine-tuning using RLHF for dialog applications.

LLM Name	Llama 2 13B GPTQ
Repository 🤗	https://huggingface.co/TheBloke/Llama-2-13B-GPTQ
Model Name	Llama 2 13B
Model Creator	Meta
Base Model(s)	Llama 2 13B Hf meta-llama/Llama-2-13b-hf
Model Size	13b
Required VRAM	7.3 GB
Updated	2026-08-07
Maintainer	TheBloke
Model Type	llama
Model Files	7.3 GB
Supported Languages	en
GPTQ Quantization	Yes
Quantization Type	gptq
Model Architecture	LlamaForCausalLM
License	llama2
Context Length	4096
Model Max Length	4096
Transformers Version	4.32.0.dev0
Tokenizer Class	LlamaTokenizer
Beginning of Sentence Token	<s>
End of Sentence Token	</s>
Unk Token	<unk>
Vocabulary Size	32000
Torch Data Type	float16

Rank the Llama 2 13B GPTQ 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 Llama 2 13B GPTQ

Best Alternatives	Context / RAM	Downloads	Likes
Yarn Llama 2 13B 128K GPTQ	128K / 7.3 GB	30	16
LongAlign 13B 64K GPTQ	64K / 7.3 GB	5	1
...boros L2 13B 2 1 YaRN 64K GPTQ	64K / 7.3 GB	6	3
Yarn Llama 2 13B 64K GPTQ	64K / 7.3 GB	20	1
OrcaMaid V3 13B 32K GPTQ	32K / 7.3 GB	13	3
OrcaMaid V2 FIX 13B 32K GPTQ	32K / 7.3 GB	12	4
EverythingLM 13B 16K GPTQ	16K / 7.3 GB	70	14
LlongOrca 13B 16K GPT	16K / 7.3 GB	5	0
Tinybra 13B GPTQ 32g 4BIT	16K / 8 GB	3	1
Tinybra 13B GPTQ 4BIT	16K / 7 GB	5	0

Note: green Score (e.g. "73.2") means that the model is better than TheBloke/Llama-2-13B-GPTQ.