camembert-ner
by Jean-Baptiste
110M params · token-classification · 120 likes · 117.4k downloads
camembert-ner is a 110M parameter model. At Q4 quantization it requires 0GB of VRAM. It runs comfortably on GeForce RTX 4090 (11932 tok/s), GeForce RTX 5090 (17894 tok/s), M4 Max 128GB (4366 tok/s).
Inference providers
| Provider | $/1M in | $/1M out | Throughput |
|---|---|---|---|
| HF Inference |
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 11932 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 17894 tok/s | comfortable |
| M4 Max 128GB | 128GB | 4366 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 2183 tok/s | comfortable |
| M4 Pro 24GB | 24GB | 2183 tok/s | comfortable |
| A100 PCIe 80 GB | 80GB | 21861 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 43971 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 10550 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 8724 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 8468 tok/s | comfortable |