roberta-base
by FacebookAI
125M params · fill-mask · 567 likes · 10.7M downloads
roberta-base is a 125M parameter model. At Q4 quantization it requires 0GB of VRAM. It runs comfortably on GeForce RTX 4090 (10529 tok/s), GeForce RTX 5090 (15790 tok/s), M4 Max 128GB (3853 tok/s).
Inference providers
| Provider | $/1M in | $/1M out | Throughput |
|---|---|---|---|
| HF Inference |
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 10529 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 15790 tok/s | comfortable |
| M4 Max 128GB | 128GB | 3853 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 1926 tok/s | comfortable |
| M4 Pro 24GB | 24GB | 1926 tok/s | comfortable |
| A100 PCIe 80 GB | 80GB | 19291 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 38800 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 9309 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 7698 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 7472 tok/s | comfortable |