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

Recurrentgemma 2B by google

 »  All LLMs  »  google  »  Recurrentgemma 2B   URL Share it on

  6489   

Recurrentgemma 2B is an open-source language model by google. Features: 2b LLM, VRAM: 10.8GB, License: gemma, HF Score: 40.4, LLM Explorer Score: 0.22, Arc: 31.4, HellaSwag: 56.9, MMLU: 34.6, TruthfulQA: 35.1, WinoGrande: 68.5, GSM8K: 16.2.

  Arxiv:1705.03551   Arxiv:1804.06876   Arxiv:1804.09301   Arxiv:1809.02789   Arxiv:1811.00937   Arxiv:1904.09728   Arxiv:1905.07830   Arxiv:1905.10044   Arxiv:1907.10641   Arxiv:1911.01547   Arxiv:1911.11641   Arxiv:2009.03300   Arxiv:2009.11462   Arxiv:2101.11718   Arxiv:2103.03874   Arxiv:2107.03374   Arxiv:2108.07732   Arxiv:2109.07958   Arxiv:2110.08193   Arxiv:2110.14168   Arxiv:2203.09509   Arxiv:2206.04615   Arxiv:2304.06364   Arxiv:2402.19427   Endpoints compatible   Recurrent gemma   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/google/recurrentgemma-2b 

Recurrentgemma 2B Benchmarks

MMLU Pro: 1.96
MUSR: 3.1
BBH: 4.82
IFEval: 30.17 vs 88 (so35)-65.7%
ARC: 31.4 vs 96.7 (so35)-67.5%
HellaSwag: 56.89 vs 95.3 (gpt4)-40.3%
MMLU: 34.61 vs 88.3 (so35)-60.8%
TruthfulQA: 35.1 vs 59 (gpt4)-40.5%
WinoGrande: 68.51 vs 87.5 (gpt4)-21.7%
GSM8K: 16.15 vs 96.4 (so35)-83.2%
MATH Lvl 5: 2.04
LLME Score: 0.22036
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Recurrentgemma 2B (google/recurrentgemma-2b)
🌟 Advertise your project 🚀

Recurrentgemma 2B Parameters and Internals

Model Type 
text generation
Use Cases 
Areas:
Research, Commercial applications
Applications:
Content creation, Text generation, Chatbots and conversational AI, Text summarization
Primary Use Cases:
Poems, Scripts, Code, Marketing copy, Email drafts
Limitations:
Bias from training data, Performance depends on context length, Language ambiguity, Factual inaccuracies, Common sense reasoning
Considerations:
Consider biases, the influence of training data, and the complexity of tasks.
Additional Notes 
RecurrentGemma is faster during inference and requires less memory compared to Gemma models.
Training Details 
Hardware Used:
TPUv5e clusters
Model Architecture:
Novel recurrent architecture
Safety Evaluation 
Methodologies:
Structured evaluations, Internal red-teaming testing
Risk Categories:
Child safety, Content safety, Representational harms, Memorization, Large-scale harms
Responsible Ai Considerations 
Fairness:
The models underwent careful scrutiny, input data pre-processing described and posterior evaluations reported in this card.
Transparency:
Details on models' architecture, capabilities, limitations, and evaluation processes shared.
Accountability:
Accountable through summarizing details in the model card.
Mitigation Strategies:
Continuous monitoring, exploration of de-biasing techniques, and education on responsible use are encouraged.
Input Output 
Input Format:
Text string (e.g., a question, a prompt, or a document to be summarized).
Accepted Modalities:
text
Output Format:
Generated English-language text in response to the input.
LLM NameRecurrentgemma 2B
Repository 🤗https://huggingface.co/google/recurrentgemma-2b 
Model Size2b
Required VRAM10.8 GB
Updated2026-04-21
Maintainergoogle
Model Typerecurrent_gemma
Model Files  5.0 GB: 1-of-3   5.0 GB: 2-of-3   0.8 GB: 3-of-3
Model ArchitectureRecurrentGemmaForCausalLM
Licensegemma
Transformers Version4.40.0.dev0
Tokenizer ClassGemmaTokenizer
Padding Token<pad>
Vocabulary Size256000
Torch Data Typefloat32

Best Alternatives to Recurrentgemma 2B

Best Alternatives
Context / RAM
Downloads
Likes
Recurrentgemma 2B It0K / 5.4 GB3066112
...ntgemma 2B It Bnb 4bit Smashed0K / 2.5 GB81

Rank the Recurrentgemma 2B 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? 53254 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.
Check out Ag3ntum — our secure, self-hosted AI agent for server management.
Release v20260328a
  Support LLM Explorer