xlm-roberta-large-xnli
by joeddav
561M params · zero-shot-classification · 285 likes · 98.1k downloads
xlm-roberta-large-xnli is a 561M parameter model. At Q4 quantization it requires 0GB of VRAM. It runs comfortably on GeForce RTX 4090 (2340 tok/s), GeForce RTX 5090 (3510 tok/s), M4 Max 128GB (856 tok/s).
Inference providers
| Provider | $/1M in | $/1M out | Throughput |
|---|---|---|---|
| HF Inference |
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 2340 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 3510 tok/s | comfortable |
| M4 Max 128GB | 128GB | 856 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 428 tok/s | comfortable |
| M4 Pro 24GB | 24GB | 428 tok/s | comfortable |
| A100 PCIe 80 GB | 80GB | 4288 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 8625 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 2069 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 1711 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 1661 tok/s | comfortable |