What are the hardware requirements for Hermes 2 Pro Llama 3 8B Marlin?

Hermes 2 Pro Llama 3 8B Marlin requires approximately 5.7 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 Hermes 2 Pro Llama 3 8B Marlin and how large is it?

Hermes 2 Pro Llama 3 8B Marlin is developed by mgoin, a model with 8b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate Hermes 2 Pro Llama 3 8B Marlin?

Hermes 2 Pro Llama 3 8B Marlin is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Hermes 2 Pro Llama 3 8B Marlin by mgoin — VRAM 5.7GB, 8K context

Name: Hermes 2 Pro Llama 3 8B Marlin
Author: mgoin

Hermes 2 Pro Llama 3 8B Marlin is an open-source language model by mgoin. Features: 8b LLM, VRAM: 5.7GB, Context: 8K, License: apache-2.0, Quantized, LLM Explorer Score: 0.12.

4-bit 4bit Axolotl Base model:nousresearch/meta-l... Base model:quantized:nousresea... Chatml Conversational Dataset:teknium/openhermes-2.5 Distillation Dpo En Endpoints compatible Finetuned Function calling Gpt4 Gptq Instruct Json mode Llama Llama-3 Quantized Region:us Rlhf Synthetic data

Model Card on HF 🤗: https://huggingface.co/mgoin/Hermes-2-Pro-Llama-3-8B-Marlin

Hermes 2 Pro Llama 3 8B Marlin Parameters and Internals

Model Type

instruct, chat, text generation

Use Cases

Areas:

research, commercial applications

Applications:

Function Calling, JSON Structured Outputs

Primary Use Cases:

General task and conversation capabilities, Function Calling

Limitations:

Limited to capabilities of datasets and training methods

Supported Languages

en (fluent)

Training Details

Data Sources:

teknium/OpenHermes-2.5, Function Calling and JSON Mode dataset

Methodology:

Retraining with updated and cleaned dataset, Function Calling and JSON Mode dataset, special system prompt, multi-turn function calling structure

Model Architecture:

Llama-3

Input Output

Input Format:

ChatML format

Accepted Modalities:

text

Output Format:

text, structured JSON outputs

Performance Tips:

Use system prompts for better performance in Function Calling.

LLM Name	Hermes 2 Pro Llama 3 8B Marlin
Repository 🤗	https://huggingface.co/mgoin/Hermes-2-Pro-Llama-3-8B-Marlin
Base Model(s)	Meta Llama 3 8B NousResearch/Meta-Llama-3-8B
Model Size	8b
Required VRAM	5.7 GB
Updated	2026-08-02
Maintainer	mgoin
Model Type	llama
Model Files	5.7 GB
Supported Languages	en
GPTQ Quantization	Yes
Quantization Type	gptq\|4bit
Model Architecture	LlamaForCausalLM
License	apache-2.0
Context Length	8192
Model Max Length	8192
Transformers Version	4.40.0
Tokenizer Class	PreTrainedTokenizerFast
Padding Token	<\|end_of_text\|>
Vocabulary Size	128288
Torch Data Type	float16

Rank the Hermes 2 Pro Llama 3 8B Marlin 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 Hermes 2 Pro Llama 3 8B Marlin

Best Alternatives	Context / RAM	Downloads	Likes
...a 3 8B Instruct 262K 4bit GPTQ	256K / 5.8 GB	5	1
... 8B Instruct 262K 4bit GPTQ 02	256K / 5.7 GB	6	0
...Instruct V0.5 GPTQ Safetensors	128K / 4.9 GB	113	0
...lama 3.1 8B Instruct GPTQ INT4	128K / 5.8 GB	9169	42
...Instruct 80K Qlora Merged GPTQ	80K / 5.8 GB	6	0
Llama3 German 8B 32K GPTQ	64K / 5.7 GB	7	0
...oLeo Instruct 8B 32K V0.1 GPTQ	64K / 5.7 GB	5	0
Tsukasa Llama 3 8B Qlora Gptq	32K / 5.8 GB	8	0
Llama 3 Soliloquy 8B V2 GPTQ	24K / 5.7 GB	10	1
Llama 3 Soliloquy 8B GPTQ	16K / 5.7 GB	10	5

Note: green Score (e.g. "73.2") means that the model is better than mgoin/Hermes-2-Pro-Llama-3-8B-Marlin.