Model Card on HF 🤗: https://huggingface.co/kyutai/helium-1-preview-2b
Helium 1 Preview 2B Benchmarks
MMLU Pro: 9.69
GPQA: 3.8
MUSR: 4.04
BBH: 10.95
IFEval: 26.14 vs 88 (so35)-70.3%
MATH Lvl 5: 1.36
LLME Score: 0.35564
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
What is the LLM Explorer Rank (Score)
Helium 1 Preview 2B Parameters and Internals
| LLM Name | Helium 1 Preview 2B |
| Repository 🤗 | https://huggingface.co/kyutai/helium-1-preview-2b |
| Model Size | 2b |
| Required VRAM | 4.3 GB |
| Updated | 2025-05-17 |
| Maintainer | kyutai |
| Model Type | helium |
| Model Files | |
| Supported Languages | en fr de it pt es |
| Model Architecture | HeliumForCausalLM |
| License | cc-by-4.0 |
| Context Length | 4096 |
| Model Max Length | 4096 |
| Transformers Version | 4.45.0.dev0 |
| Tokenizer Class | PreTrainedTokenizerFast |
| Vocabulary Size | 48000 |
| Torch Data Type | bfloat16 |
Rank the Helium 1 Preview 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! 🌟
What open-source LLMs or SLMs are you in search of? 47402 in total.
Original data from HuggingFace, OpenCompass and various public git repos.
Release v20241227