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

Smol Llama 4x220M MoE by Isotonic

 »  All LLMs  »  Isotonic  »  Smol Llama 4x220M MoE   URL Share it on

  103   
  Autotrain compatible Bee-spoke-data/beecoder-220m-p... Bee-spoke-data/smol llama-220m... Bee-spoke-data/zephyr-220m-dpo... Bee-spoke-data/zephyr-220m-sft... Dataset:bigcode/the-stack-smol... Dataset:eleutherai/proof-pile-... Dataset:huggingfaceh4/ultracha... Dataset:huggingfaceh4/ultrafee...   Dataset:jeankaddour/minipile Dataset:mattymchen/refinedweb-... Dataset:pszemraj/simple wikipe...   Dataset:teknium/openhermes   Endpoints compatible   Lazymergekit   Merge   Mergekit   Mixtral   Moe   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/Isotonic/smol_llama-4x220M-MoE 

Smol Llama 4x220M MoE Benchmarks

ARC: 25.09 vs 96.7 (so35)-74.1%
HellaSwag: 29.24 vs 95.3 (gpt4)-69.3%
MMLU: 25.88 vs 88.3 (so35)-70.7%
TruthfulQA: 43.92 vs 59 (gpt4)-25.6%
WinoGrande: 51.22 vs 87.5 (gpt4)-41.5%
GSM8K: 0.15 vs 96.4 (so35)-99.8%
LLME Score: 0.21068
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Smol Llama 4x220M MoE (Isotonic/smol_llama-4x220M-MoE)

Smol Llama 4x220M MoE Parameters and Internals

Model Type 
Mixure of Experts (MoE), text-generation
Training Details 
Data Sources:
JeanKaddour/minipile, pszemraj/simple_wikipedia_LM, mattymchen/refinedweb-3m, HuggingFaceH4/ultrachat_200k, teknium/openhermes, HuggingFaceH4/ultrafeedback_binarized, EleutherAI/proof-pile-2, bigcode/the-stack-smol-xl
Methodology:
LazyMergekit was used to create the MoE model with specific source models and their positive prompts.
LLM NameSmol Llama 4x220M MoE
Repository 🤗https://huggingface.co/Isotonic/smol_llama-4x220M-MoE 
Model Size595.4m
Required VRAM1.2 GB
Updated2025-03-14
MaintainerIsotonic
Model Typemixtral
Model Files  1.2 GB: 1-of-1
Model ArchitectureMixtralForCausalLM
Licenseapache-2.0
Context Length2048
Model Max Length2048
Transformers Version4.37.2
Tokenizer ClassLlamaTokenizer
Padding Token<s>
Vocabulary Size32128
Torch Data Typebfloat16

Best Alternatives to Smol Llama 4x220M MoE

Best Alternatives
Context / RAM
Downloads
Likes
...inyMixtral 4x220M UniversalNER2K / 2.4 GB180

Rank the Smol Llama 4x220M MoE 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? 45019 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