What are the hardware requirements for Llama 3 15B Instruct Zeroed?

Llama 3 15B Instruct Zeroed requires approximately 30 GB of VRAM and supports a context window of 8K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Llama 3 15B Instruct Zeroed and how large is it?

Llama 3 15B Instruct Zeroed is developed by elinas, a model with 8b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does Llama 3 15B Instruct Zeroed perform on standard benchmarks?

Llama 3 15B Instruct Zeroed has the following published scores: MMLU 67.97. Compare against reference models on this page or on the LLM Explorer leaderboards.

Llama 3 15B Instruct Zeroed by elinas — VRAM 30GB, 8K context

Name: Llama 3 15B Instruct Zeroed
Author: elinas

Llama 3 15B Instruct Zeroed is an open-source language model by elinas. Features: 8b LLM, VRAM: 30GB, Context: 8K, License: llama3, Instruction-Based, Merged, LLM Explorer Score: 0.14, Arc: 61.7, HellaSwag: 78.6, MMLU: 68, GSM8K: 70.7.

Merged Model Base model:finetune:meta-llama... Base model:meta-llama/meta-lla... Conversational Endpoints compatible Instruct Llama Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/elinas/Llama-3-15B-Instruct-zeroed

Llama 3 15B Instruct Zeroed Parameters and Internals

Model Type

text generation

Additional Notes

This model was merged using the passthrough merge method while zeroing `o_proj` and `down_proj`. There is a finetuned version available that improves performance.

LLM Name	Llama 3 15B Instruct Zeroed
Repository 🤗	https://huggingface.co/elinas/Llama-3-15B-Instruct-zeroed
Base Model(s)	Meta Llama 3 8B Instruct meta-llama/Meta-Llama-3-8B-Instruct
Merged Model	Yes
Model Size	8b
Required VRAM	30 GB
Updated	2026-07-18
Maintainer	elinas
Model Type	llama
Instruction-Based	Yes
Model Files	10.0 GB: 1-of-4 9.9 GB: 2-of-4 10.0 GB: 3-of-4 0.1 GB: 4-of-4
Model Architecture	LlamaForCausalLM
License	llama3
Context Length	8192
Model Max Length	8192
Transformers Version	4.37.1
Tokenizer Class	PreTrainedTokenizerFast
Vocabulary Size	128256
Torch Data Type	bfloat16

Rank the Llama 3 15B Instruct Zeroed 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

Quantized Models of the Llama 3 15B Instruct Zeroed

Model	Likes	Downloads	VRAM
L3 Aethora 15B V2 EXL2 6.0bpw	1	6	11 GB
L3 Aethora 15B V2 EXL2 8.0bpw	1	4	12 GB
L3 Aethora 15B V2 EXL2 4.0bpw	1	2	8 GB
L3 Aethora 15B V2 EXL2 5.0bpw	1	0	10 GB

Best Alternatives to Llama 3 15B Instruct Zeroed

Best Alternatives	Context / RAM	Downloads	Likes
...otron 8B UltraLong 4M Instruct	4192K / 32.1 GB	1135	125
UltraLong Thinking	4192K / 16.1 GB	2	3
...a 3.1 8B UltraLong 4M Instruct	4192K / 32.1 GB	176	24
...a 3.1 8B UltraLong 2M Instruct	2096K / 32.1 GB	875	9
...otron 8B UltraLong 2M Instruct	2096K / 32.1 GB	124	18
Cthulhu 8B V1.4	1048K / 16.1 GB	10	10
...raLong 1M Instruct Abliterated	1048K / 32.1 GB	4	9
...a 3.1 8B UltraLong 1M Instruct	1048K / 32.1 GB	1387	29
...otron 8B UltraLong 1M Instruct	1048K / 32.1 GB	702	59
Zero Llama 3.1 8B Beta6	1048K / 16.1 GB	7	1

Note: green Score (e.g. "73.2") means that the model is better than elinas/Llama-3-15B-Instruct-zeroed.