LLM EXPLORER
 LLM Explorer: A Curated Large Language Model Directory and Analytics
Dark Theme
  LLM  List    LLM Hosting    LLM Leaderboards    Blog    Newsfeed    Advertise

Ahma 7B by Finnish-NLP

 »  All LLMs  »  Finnish-NLP  »  Ahma 7B   URL Share it on

  84   
  Arxiv:2302.06675   Arxiv:2302.13971   Arxiv:2305.16264   Autotrain compatible   Conversational Dataset:finnish-nlp/culturax f... Dataset:finnish-nlp/hplt 1.2 f... Dataset:finnish-nlp/reddit fi ... Dataset:finnish-nlp/wikipedia ... Dataset:intfloat/multilingual ...   Doi:10.57967/hf/4012   Fi   Finnish   Llama   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/Finnish-NLP/Ahma-7B 

Ahma 7B Benchmarks

LLME Score: 0.15591
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Ahma 7B (Finnish-NLP/Ahma-7B)

Ahma 7B Parameters and Internals

Model Type 
text generation
Use Cases 
Areas:
research, text generation
Applications:
text generation
Primary Use Cases:
text generation, instruction following
Limitations:
Not suitable for multilingual, Not suitable for code generation
Additional Notes 
Training data was filtered with a toxicity classifier. The model is case-sensitive and differentiates between 'finnish' and 'Finnish'. Known for its strong non-arithmetic task performance in 0-shot settings.
Supported Languages 
Finnish (Proficient)
Training Details 
Data Sources:
Finnish-NLP/CulturaX_fi_cleaned, Finnish-NLP/HPLT_1.2_fi_cleaned, Finnish-NLP/wikipedia_20231101_fi_cleaned, Finnish-NLP/Reddit_fi_2006_2022, Yle Finnish News Archive 2011-2018, Yle Finnish News Archive 2019-2020, Finnish News Agency Archive (STT), The Suomi24 Sentences Corpus, Project Lönnrot, Finnish parliament speeches, intfloat/multilingual_cc_news, fi-news-corpus, Finnish higher education public theses, Finnish single-turn instruction-following datasets
Data Volume:
41 billion tokens
Methodology:
2-stage pretraining with instruction-following examples
Context Length:
2048
Hardware Used:
TPUv4-32 VM
Model Architecture:
Decoder-only transformer
Input Output 
Input Format:
Byte Pair Encoding (BPE) using SentencePiece
Accepted Modalities:
text
Output Format:
Text
LLM NameAhma 7B
Repository 🤗https://huggingface.co/Finnish-NLP/Ahma-7B 
Model Size7b
Required VRAM14 GB
Updated2025-05-09
MaintainerFinnish-NLP
Model Typellama
Model Files  5.0 GB: 1-of-3   5.0 GB: 2-of-3   4.0 GB: 3-of-3
Supported Languagesfi
Model ArchitectureLlamaForCausalLM
Licenseapache-2.0
Context Length2048
Model Max Length2048
Transformers Version4.38.0.dev0
Tokenizer ClassPreTrainedTokenizerFast
Vocabulary Size64256
Torch Data Typefloat16

Best Alternatives to Ahma 7B

Best Alternatives
Context / RAM
Downloads
Likes
A6 L1024K / 16.1 GB2010
M1024K / 16.1 GB1270
1571024K / 16.1 GB1010
1241024K / 16.1 GB930
A3.41024K / 16.1 GB130
A5.41024K / 16.1 GB120
A2.41024K / 16.1 GB120
2 Very Sci Fi1024K / 16.1 GB3170
1621024K / 16.1 GB600
1181024K / 16.1 GB150
Note: green Score (e.g. "73.2") means that the model is better than Finnish-NLP/Ahma-7B.

Rank the Ahma 7B 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? 47215 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.
Release v20241227
  Support LLM Explorer