What are the hardware requirements for MN 12B Lyra V3?

MN 12B Lyra V3 requires approximately 24.4 GB of VRAM and supports a context window of 1000K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed MN 12B Lyra V3 and how large is it?

MN 12B Lyra V3 is developed by Sao10K, a model with 12b 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 MN 12B Lyra V3?

MN 12B Lyra V3 is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

MN 12B Lyra V3 by Sao10K — VRAM 24.4GB, 1000K context

Name: MN 12B Lyra V3
Author: Sao10K

MN 12B Lyra V3 is an open-source language model by Sao10K. Features: 12b LLM, VRAM: 24.4GB, Context: 1000K, License: proprietary, LLM Explorer Score: 0.19.

En Mistral Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/Sao10K/MN-12B-Lyra-v3

MN 12B Lyra V3 Parameters and Internals

Model Type

text generation

Additional Notes

Original format was meant for an 8B pruned NeMo train.

Training Details

Data Sources:

A custom, cleaned mix of Stheno-v3.4's Dataset

Methodology:

Multi-turn training with RL step, Magpie-style responses.

Training Time:

6 Hours

Hardware Used:

4xH100 SXM

Input Output

Input Format:

ChatML-style prompting Format

Accepted Modalities:

text

Output Format:

text

Performance Tips:

Recommended samplers: Temperature: 0.7 - 1.2, min_p: 0.1 - 0.2

LLM Name	MN 12B Lyra V3
Repository 🤗	https://huggingface.co/Sao10K/MN-12B-Lyra-v3
Model Size	12b
Required VRAM	24.4 GB
Updated	2024-11-27
Maintainer	Sao10K
Model Type	mistral
Model Files	2.0 GB: 1-of-13 1.9 GB: 2-of-13 1.9 GB: 3-of-13 1.9 GB: 4-of-13 1.9 GB: 5-of-13 1.9 GB: 6-of-13 1.9 GB: 7-of-13 1.9 GB: 8-of-13 1.9 GB: 9-of-13 1.9 GB: 10-of-13 1.9 GB: 11-of-13 1.9 GB: 12-of-13 1.5 GB: 13-of-13
Supported Languages	en
Gated Model	Yes
Model Architecture	MistralForCausalLM
License	proprietary
Context Length	1024000
Model Max Length	1024000
Transformers Version	4.43.4
Tokenizer Class	PreTrainedTokenizerFast
Padding Token	[/INST]
Vocabulary Size	131072
Torch Data Type	bfloat16

Rank the MN 12B Lyra V3 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 MN 12B Lyra V3

Best Alternatives	Context / RAM	Downloads	Likes
...r Nemo 12B Instruct R 21 09 24	1000K / 24.5 GB	396277	141
MN 12B Mag Mell R1	1000K / 24.5 GB	69558	238
Riverfish Rocinante 12B SFT	1000K / 49.4 GB	166	1
Riverfish Rocinante 12B	1000K / 49.4 GB	111	1
MN 12B LucidFaun RP RU	1000K / 24.5 GB	1005	6
Qliphoth 12B V1.2	1000K / 24.5 GB	107	7
Captain Eris BMO Violent 12B	1000K / 24.5 GB	305	2
Dans SakuraKaze V1.0.0 12B	1000K / 24.5 GB	7	21
...n Eris BMO Violent GRPO V0.420	1000K / 24.5 GB	92	3
Shining Gold Sakura 12B	1000K / 24.5 GB	52	2

Note: green Score (e.g. "73.2") means that the model is better than Sao10K/MN-12B-Lyra-v3.