LLM EXPLORER 65,548 MODELS INDEXED

Nous Hermes 2 Mixtral 8x7B DPO 5.0bpw H6 EXL2 by LoneStriker

By LoneStriker · 12 downloads
HF Repo ↗ Copy URL

Nous Hermes 2 Mixtral 8x7B DPO 5.0bpw H6 EXL2 is an open-source language model by LoneStriker. Features: LLM, VRAM: 29.5GB, Context: 32K, License: apache-2.0, MoE, Quantized, LLM Explorer Score: 0.1.

Base model:finetune:mistralai/... Base model:mistralai/mixtral-8...   Chatml   Conversational   Distillation   Dpo   En   Endpoints compatible   Exl2   Finetuned   Gpt4   Instruct   Mixtral   Moe   Quantized   Region:us   Rlhf   Safetensors   Sharded   Synthetic data   Tensorflow
Model Card on HF 🤗: https://huggingface.co/LoneStriker/Nous-Hermes-2-Mixtral-8x7B-DPO-5.0bpw-h6-exl2 

Nous Hermes 2 Mixtral 8x7B DPO 5.0bpw H6 EXL2 Parameters and Internals

Model Type 
instruct, finetune, chatml, DPO, RLHF, gpt4, synthetic data, distillation
Additional Notes 
The model was developed using sponsorship from Together.ai.
Supported Languages 
en (Proficient)
Training Details 
Data Sources:
GPT-4 generated data, Other high quality open datasets
Data Volume:
1,000,000+ entries
Methodology:
SFT + DPO
Hardware Used:
Sponsored by Together.ai
Input Output 
Input Format:
ChatML with roles for turns and system prompt capability.
Accepted Modalities:
text
Output Format:
Text in chat dialogue format with roles and turns.
Performance Tips:
Ensure to use the ChatML format to guide interaction with appropriate messages and responses.
LLM NameNous Hermes 2 Mixtral 8x7B DPO 5.0bpw H6 EXL2
Repository 🤗https://huggingface.co/LoneStriker/Nous-Hermes-2-Mixtral-8x7B-DPO-5.0bpw-h6-exl2 
Base Model(s)  mistralai/Mixtral-8x7B-v0.1   mistralai/Mixtral-8x7B-v0.1
Required VRAM29.5 GB
Updated2026-07-18
MaintainerLoneStriker
Model Typemixtral
Model Files  8.6 GB: 1-of-4   8.6 GB: 2-of-4   8.6 GB: 3-of-4   3.7 GB: 4-of-4
Supported Languagesen
Quantization Typeexl2
Model ArchitectureMixtralForCausalLM
Licenseapache-2.0
Context Length32768
Model Max Length32768
Transformers Version4.37.0.dev0
Tokenizer ClassLlamaTokenizer
Padding Token</s>
Vocabulary Size32002
Torch Data Typebfloat16

Best Alternatives to Nous Hermes 2 Mixtral 8x7B DPO 5.0bpw H6 EXL2

Best Alternatives
Context / RAM
Downloads
Likes
...oE V0.1 DPO F16 4.0bpw H6 EXL2195K / 31.3 GB70
...oE V0.1 DPO F16 5.0bpw H6 EXL2195K / 38.8 GB60
...2 Mixtral 8x22b 6.0bpw H8 EXL264K / 105.8 GB21
WizardLM 2 8x22 EXL2 4.0bpw64K / 70.9 GB71
...M 2 8x22B Beige 5.0bpw H6 EXL264K / 88.5 GB80
...M 2 8x22B Beige 4.0bpw H6 EXL264K / 70.8 GB60
...M 2 8x22B Beige 2.4bpw H6 EXL264K / 42.7 GB50
...M 2 8x22B Beige 3.0bpw H6 EXL264K / 53.2 GB50
...rdLM 2 8x22B Beige EXL2 5.0bpw64K / 88.4 GB50
...B Instruct V0.1 8.0bpw H8 EXL264K / 120.2 GB61
Note: green Score (e.g. "73.2") means that the model is better than LoneStriker/Nous-Hermes-2-Mixtral-8x7B-DPO-5.0bpw-h6-exl2.