What are the hardware requirements for Mistral 7B Instruct V0.2 2x7B MoE?

Mistral 7B Instruct V0.2 2x7B MoE requires approximately 25.8 GB of VRAM and supports a context window of 32K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Mistral 7B Instruct V0.2 2x7B MoE and how large is it?

Mistral 7B Instruct V0.2 2x7B MoE is developed by perlthoughts, a model with 12.9b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does Mistral 7B Instruct V0.2 2x7B MoE perform on standard benchmarks?

Mistral 7B Instruct V0.2 2x7B MoE has the following published scores: MMLU 60.74. Compare against reference models on this page or on the LLM Explorer leaderboards.

Mistral 7B Instruct V0.2 2x7B MoE by perlthoughts — VRAM 25.8GB, 32K context

Name: Mistral 7B Instruct V0.2 2x7B MoE
Author: perlthoughts

Mistral 7B Instruct V0.2 2x7B MoE is an open-source language model by perlthoughts. Features: 12.9b LLM, VRAM: 25.8GB, Context: 32K, License: apache-2.0, MoE, Fine-Tuned, Instruction-Based, Merged, LLM Explorer Score: 0.13, Arc: 63, HellaSwag: 84.9, MMLU: 60.7, GSM8K: 39.4.

Merged Model Arxiv:2310.06825 Autotrain compatible Conversational Finetuned Instruct License:apache-2.0 Mixtral Model-index Moe Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/perlthoughts/Mistral-7B-Instruct-v0.2-2x7B-MoE

Mistral 7B Instruct V0.2 2x7B MoE Parameters and Internals

LLM Name	Mistral 7B Instruct V0.2 2x7B MoE
Repository 🤗	https://huggingface.co/perlthoughts/Mistral-7B-Instruct-v0.2-2x7B-MoE
Merged Model	Yes
Model Size	12.9b
Required VRAM	25.8 GB
Updated	2024-07-15
Maintainer	perlthoughts
Model Type	mixtral
Instruction-Based	Yes
Model Files	5.0 GB: 1-of-6 4.9 GB: 2-of-6 5.0 GB: 3-of-6 5.0 GB: 4-of-6 4.9 GB: 5-of-6 1.0 GB: 6-of-6
Model Architecture	MixtralForCausalLM
License	apache-2.0
Context Length	32768
Model Max Length	32768
Transformers Version	4.37.0.dev0
Tokenizer Class	LlamaTokenizer
Padding Token	<s>
Vocabulary Size	32000
Torch Data Type	float16

Rank the Mistral 7B Instruct V0.2 2x7B MoE 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 Mistral 7B Instruct V0.2 2x7B MoE

Best Alternatives	Context / RAM	Downloads	Likes
Inf Silent Kunoichi V0.1 2x7B	32K / 25.6 GB	8	0
Inf Silent Kunoichi V0.2 2x7B	32K / 25.6 GB	9	1
NearalMistral 2x7B	32K / 25.8 GB	50	1
Megatron V3 2x7B	32K / 25.8 GB	76	3
MistarlingMaid 2x7B Base	32K / 25.8 GB	52	0
MergedExpert 2x8b	32K / 25.8 GB	5	0
MergedExperts 2x8b	32K / 25.8 GB	6	0
...afted Hermetic Platypus C 2x7B	32K / 25.8 GB	49	0
...tral Instruct MoE Experimental	32K / 25.8 GB	56	2
Mistral 2x7b V0.1	32K / 25.8 GB	6	1

Note: green Score (e.g. "73.2") means that the model is better than perlthoughts/Mistral-7B-Instruct-v0.2-2x7B-MoE.