Compare LLM Models
Select 2-5 models to compare side by side.
Select models to compare
0 / 5 selected
Top downloads
Select 2-5 models
| Attribute |
01-ai
|
|---|---|
| Company | 01-ai |
| Release Date | 2024-03-16 |
| Parameters (B) | 9 |
| Architecture | Transformer |
| Context Window | 200000 |
| Input Modalities | text |
| Open Source | Yes |
| License | apache-2.0 |
| Score | 0 |
| VRAM (GB) | 50 |
| Compute | BF16 weights need ~20 GB VRAM (1× RTX 4090 24 GB); full 200K context inference needs ~50 GB minimum (1× A800 80 GB recommended, per Yi-6B-200K deployment guidance on the HuggingFace model card) |
| Benchmark: ARC | 57.6 |
| Benchmark: BBH | 47.8 |
| Benchmark: GPQA-Diamond | 29.8 |
| Benchmark: GSM8K | 52.3 |
| Benchmark: HellaSwag | 76.4 |
| Benchmark: HumanEval | 39 |
| Benchmark: MATH | 15.9 |
| Benchmark: MBPP | 54.4 |
| Benchmark: MMLU | 68.4 |
| Benchmark: MMLU-Pro | 36.2 |
| Benchmark: TruthfulQA | 33.8 |
| Pricing (Input) | — |
| Pricing (Output) | — |