Who developed Aira 2 355M and how large is it?

Aira 2 355M is developed by nicholasKluge, a model with 355m parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does Aira 2 355M perform on standard benchmarks?

Aira 2 355M has the following published scores: MMLU 27.26. Compare against reference models on this page or on the LLM Explorer leaderboards.

Aira 2 355M by nicholasKluge

Name: Aira 2 355M
Author: nicholasKluge

Aira 2 355M is an open-source language model by nicholasKluge. Features: 355m LLM, License: apache-2.0, Instruction-Based, LLM Explorer Score: 0.12, Arc: 27.6, HellaSwag: 38.9, MMLU: 27.3.

Arxiv:1803.05457 Arxiv:2109.07958 Arxiv:2203.09509 Alignment Assistant Base model:finetune:openai-com... Base model:openai-community/gp... Co2 eq emissions Conversation Dataset:nicholaskluge/instruct... En Endpoints compatible Gpt2 Instruct Pytorch Region:us Safetensors

Model Card on HF 🤗: https://huggingface.co/nicholasKluge/Aira-2-355M

Aira 2 355M Parameters and Internals

Model Type

text generation, instruction tuned, conversation, assistant

Use Cases

Areas:

research, commercial applications

Applications:

instruction following, text generation, virtual assistance

Primary Use Cases:

interactive dialogues, instruction response

Limitations:

May produce hallucinations, biases, toxic content, Repetition and verbosity issues

Supported Languages

English (Primary)

Training Details

Data Sources:

Instruct-Aira Dataset

Methodology:

fine-tuning

Hardware Used:

1 NVIDIA A100-SXM4-40GB

Input Output

Input Format:

Use special tokens <|startofinstruction|>, <|endofinstruction|>

Accepted Modalities:

text

Output Format:

Generated text response

Performance Tips:

Avoid repetition penalty set to a low value to prevent loops

LLM Name	Aira 2 355M
Repository 🤗	https://huggingface.co/nicholasKluge/Aira-2-355M
Base Model(s)	gpt2-medium /gpt2-medium
Model Size	355m
Required VRAM	0 GB
Updated	2026-07-16
Maintainer	nicholasKluge
Model Type	gpt2
Instruction-Based	Yes
Model Files	1.4 GB 1.0 GB 1.4 GB 0.0 GB 0.0 GB
Supported Languages	en
Model Architecture	GPT2LMHeadModel
License	apache-2.0
Model Max Length	1024
Transformers Version	4.33.1
Tokenizer Class	GPT2Tokenizer
Beginning of Sentence Token	<\|startofinstruction\|>
End of Sentence Token	<\|endofcompletion\|>
Unk Token	<\|endoftext\|>
Vocabulary Size	50261
Torch Data Type	float32
Activation Function	gelu_new
Errors	replace

Rank the Aira 2 355M 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 Aira 2 355M

Best Alternatives	Context / RAM	Downloads	Likes
SSH 355M	0K / 1.4 GB	104	0

Note: green Score (e.g. "73.2") means that the model is better than nicholasKluge/Aira-2-355M.