facial_emotions_image_detection
by dima806
86M params · image-classification · 120 likes · 71.4k downloads
facial_emotions_image_detection is a 86M parameter model. At Q4 quantization it requires 0GB of VRAM. It runs comfortably on GeForce RTX 4090 (15302 tok/s), GeForce RTX 5090 (22947 tok/s), M4 Max 128GB (5599 tok/s).
Inference providers
| Provider | $/1M in | $/1M out | Throughput |
|---|---|---|---|
| HF Inference |
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 15302 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 22947 tok/s | comfortable |
| M4 Max 128GB | 128GB | 5599 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 2799 tok/s | comfortable |
| M4 Pro 24GB | 24GB | 2799 tok/s | comfortable |
| A100 PCIe 80 GB | 80GB | 28035 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 56388 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 13529 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 11188 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 10860 tok/s | comfortable |