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

Falcon2 5.5B Multilingual by ssmits

 »  All LLMs  »  ssmits  »  Falcon2 5.5B Multilingual   URL Share it on

  412   
  Merged Model   Autotrain compatible Base model:finetune:tiiuae/fal...   Base model:tiiuae/falcon-11b   Conversational   Cs   Custom code   Da   De   Endpoints compatible   Es   Falcon   Fr   It   Nl   No   Pl   Pt   Region:us   Ro   Safetensors   Sharded   Sv   Tensorflow   Tiiuae/falcon-11b
Model Card on HF 🤗: https://huggingface.co/ssmits/Falcon2-5.5B-multilingual 

Falcon2 5.5B Multilingual Benchmarks

ARC: 26.11 vs 96.7 (so35)-73%
HellaSwag: 40.3 vs 95.3 (gpt4)-57.7%
MMLU: 23.91 vs 88.3 (so35)-72.9%
TruthfulQA: 47.56 vs 59 (gpt4)-19.4%
WinoGrande: 55.25 vs 87.5 (gpt4)-36.9%
LLME Score: 0.26058
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Falcon2 5.5B Multilingual (ssmits/Falcon2-5.5B-multilingual)

Falcon2 5.5B Multilingual Parameters and Internals

Model Type 
text generation
Use Cases 
Areas:
research, specialization, fine-tuning
Applications:
summarization, text generation, chatbot
Primary Use Cases:
text generation across supported languages
Limitations:
Limited generalization to languages outside the trained set.
Considerations:
Appropriate precautions for production uses.
Additional Notes 
Ensure evaluation of harm and biases for any production deployments.
Supported Languages 
es (fluent), fr (fluent), de (fluent), no (fluent), sv (fluent), da (fluent), nl (fluent), pt (fluent), pl (fluent), ro (fluent), it (fluent), cs (fluent)
Training Details 
Data Sources:
wikimedia/wikipedia subsets of 11 languages
Data Volume:
5 trillion tokens
Methodology:
Pruning using PruneMe with analysis across multiple languages
Model Architecture:
Transformed from Falcon-11B using passthrough merge method
Safety Evaluation 
Methodologies:
layer similarity analysis
Findings:
model carries typical online stereotypes and biases
Risk Categories:
bias, generalization
Ethical Considerations:
Model trained on large-scale, web-representative corpora; potential presence of biases.
Responsible Ai Considerations 
Fairness:
Ensure model deployment evaluates fairness and bias.
Transparency:
Pruning methodolgy is documented, but not easy to reverse.
Accountability:
Deploying organization should be accountable for harm from outputs.
Mitigation Strategies:
Finetuning and guardrails recommended.
Input Output 
Input Format:
text input in any of the supported languages.
Accepted Modalities:
text
Output Format:
generated text based on input prompt.
Performance Tips:
Fine-tuning recommended for specific domain applications.
LLM NameFalcon2 5.5B Multilingual
Repository 🤗https://huggingface.co/ssmits/Falcon2-5.5B-multilingual 
Base Model(s)  Falcon 11B   tiiuae/falcon-11B
Merged ModelYes
Model Size11b
Required VRAM10.9 GB
Updated2024-12-22
Maintainerssmits
Model Typefalcon
Model Files  0.9 GB: 1-of-12   1.0 GB: 2-of-12   0.9 GB: 3-of-12   0.9 GB: 4-of-12   1.0 GB: 5-of-12   0.9 GB: 6-of-12   0.9 GB: 7-of-12   1.0 GB: 8-of-12   0.9 GB: 9-of-12   0.9 GB: 10-of-12   1.0 GB: 11-of-12   0.6 GB: 12-of-12
Supported Languageses fr de no sv da nl pt pl ro it cs
Model ArchitectureFalconForCausalLM
Licenseapache-2.0
Context Length8192
Model Max Length8192
Transformers Version4.40.2
Is Biased0
Tokenizer ClassPreTrainedTokenizerFast
Padding Token<|endoftext|>
Vocabulary Size65024
Torch Data Typebfloat16

Best Alternatives to Falcon2 5.5B Multilingual

Best Alternatives
Context / RAM
Downloads
Likes
Falcon 11B8K / 22.1 GB21564213
Falcon2 5.5B Polish8K / 10.9 GB21961
Falcon2 5.5B German8K / 10.9 GB3980
Falcon2 11B8K / 6.6 GB180
Enron Falcon 11B8K / 7.6 GB141
Falcon2 5.5B Czech8K / 10.9 GB420
Falcon2 5.5B Portuguese8K / 10.9 GB270
Falcon2 5.5B Spanish8K / 10.9 GB220
Falcon2 5.5B Norwegian8K / 10.9 GB211
Falcon2 5.5B Dutch8K / 10.9 GB291
Note: green Score (e.g. "73.2") means that the model is better than ssmits/Falcon2-5.5B-multilingual.

Rank the Falcon2 5.5B Multilingual 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? 40066 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 v20241217
  Support LLM Explorer