segformer_b2_clothes
by mattmdjaga
27M params · image-segmentation · 488 likes · 264.3k downloads
segformer_b2_clothes is a 27M parameter model. At Q4 quantization it requires 0GB of VRAM. It runs comfortably on GeForce RTX 4090 (47986 tok/s), GeForce RTX 5090 (71961 tok/s), M4 Max 128GB (17560 tok/s).
Inference providers
| Provider | $/1M in | $/1M out | Throughput |
|---|---|---|---|
| HF Inference |
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 47986 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 71961 tok/s | comfortable |
| M4 Max 128GB | 128GB | 17560 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 8780 tok/s | comfortable |
| M4 Pro 24GB | 24GB | 8780 tok/s | comfortable |
| A100 PCIe 80 GB | 80GB | 87917 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 176829 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 42427 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 35085 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 34055 tok/s | comfortable |