Name: Mpt 7B Instruct
Author: mosaicml

Mpt 7B Instruct is an open-source language model by mosaicml. Features: 7b LLM, VRAM: 13.3GB, License: apache-2.0, Instruction-Based, HF Score: 44.8, LLM Explorer Score: 0.16, Arc: 38.4, HellaSwag: 75.8, MMLU: 30.6, TruthfulQA: 35.1, WinoGrande: 70.5, GSM8K: 7.8.

Arxiv:2010.04245 Arxiv:2108.12409 Arxiv:2205.14135 Autotrain compatible Composer Custom code Dataset:mosaicml/dolly hhrlhf Instruct Llm-foundry Mosaicml Mpt Pytorch Region:us Sharded

Model Card on HF 🤗: https://huggingface.co/mosaicml/mpt-7b-instruct

Mpt 7B Instruct Benchmarks

ARC: 38.4 vs 96.7 (so35)^-60.3%

HellaSwag: 75.8 vs 95.3 (gpt4)^-20.5%

MMLU: 30.6 vs 88.3 (so35)^-65.3%

TruthfulQA: 35.08 vs 59 (gpt4)^-40.5%

WinoGrande: 70.48 vs 87.5 (gpt4)^-19.5%

GSM8K: 7.8 vs 96.4 (so35)^-91.9%

OpenCompas: 36.8 vs 73.3 (gpt4)^-49.8%

Exam: 30.2 vs 77.2 (gpt4)^-60.9%

Language: 32 vs 62 (gpt4)^-48.4%

Knowledge: 40.8 vs 73.5 (gpt4)^-44.5%

Understand: 37.5 vs 70 (gpt4)^-46.4%

Reasoning: 38.7 vs 74.4 (gpt4)^-48%

LLME Score: 0.1616

^nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

What is the LLM Explorer Rank (Score)

Mpt 7B Instruct (mosaicml/mpt-7b-instruct)

🌟 Advertise your project 🚀

Mpt 7B Instruct Parameters and Internals

Model Type

Instruction following, Text generation

Use Cases

Areas:

Research, Commercial applications

Primary Use Cases:

Short-form instruction following

Limitations:

Can produce factually incorrect or offensive outputs

Considerations:

Should not be relied on to produce factually accurate information

Additional Notes

Uses a custom MPT architecture and requires `trust_remote_code=True` in some processes.

Training Details

Data Sources:

databricks/databricks-dolly-15k, Anthropic/hh-rlhf

Methodology:

Decoder-only transformer architecture with modifications such as FlashAttention, ALiBi, no positional embeddings, no biases

Context Length:

2048

Training Time:

2.3 hours

Hardware Used:

8 A100-40GB GPUs

Model Architecture:

Modified decoder-only transformer with 32 layers and 32 heads, d_model: 4096, vocab size: 50432

Input Output

Input Format:

Formatted prompts using instruction and response keys from dolly-15k format

Accepted Modalities:

Text

Output Format:

Textual response

Performance Tips:

Use torch.autocast for lower precision runs. Increase max sequence length with ALiBi.

LLM Name	Mpt 7B Instruct
Repository 🤗	https://huggingface.co/mosaicml/mpt-7b-instruct
Model Size	7b
Required VRAM	13.3 GB
Updated	2025-09-23
Maintainer	mosaicml
Model Type	mpt
Instruction-Based	Yes
Model Files	9.9 GB: 1-of-2 3.4 GB: 2-of-2
Model Architecture	MPTForCausalLM
License	apache-2.0
Model Max Length	2048
Transformers Version	4.28.1
Tokenizer Class	GPTNeoXTokenizer
Vocabulary Size	50432
Torch Data Type	bfloat16

Quantized Models of the Mpt 7B Instruct

Model	Likes	Downloads	VRAM
Mpt 7B Instruct Q8	2	6	6 GB

Best Alternatives to Mpt 7B Instruct

Best Alternatives	Context / RAM	Downloads	Likes
Mpt 7B Chat	0K / 13.3 GB	80920	518
Mpt 7B Int8 Ov	0K / 0 GB	13	0
Mpt 7B 8K Instruct	0K / 13.3 GB	2012	27
Sea Lion 7B Instruct	0K / 15 GB	208	23
Sea Lion 7B Instruct Research	0K / 15 GB	11	14
Results	0K / 13.3 GB	7	0
Mpt 7B 8K Chat Sharded Bf16	0K / 13.4 GB	4	1
...7B 8K Instruct Peft Compatible	0K / 13.3 GB	0	1
Vigogne Mpt 7B Instruct	0K / 13.4 GB	5	0
Mpt 7B Instruct	0K / 13.3 GB	1092	1

Note: green Score (e.g. "73.2") means that the model is better than mosaicml/mpt-7b-instruct.

Rank the Mpt 7B Instruct Capabilities

🆘 Have you tried this model? Rate its performance. This feedback would greatly assist ML community in identifying the most suitable model for their needs. Your contribution really does make a difference! 🌟

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

What open-source LLMs or SLMs are you in search of? 52721 in total.

Email us: info@extractum.io. Our Privacy Policy | Terms and Conditions | Suggest an improvement.

Our Social Media →

Original data from HuggingFace, OpenCompass and various public git repos.

Check out Ag3ntum — our secure, self-hosted AI agent for server management.

Release v20260328a

Support LLM Explorer

Mpt 7B Instruct by mosaicml

» All LLMs » mosaicml » Mpt 7B Instruct URL Share it on