text2vec-base-chinese
by shibing624
102M params · sentence-similarity · 784 likes · 409.8k downloads
text2vec-base-chinese is a 102M parameter model. At Q4 quantization it requires 0GB of VRAM. It runs comfortably on GeForce RTX 4090 (12838 tok/s), GeForce RTX 5090 (19253 tok/s), M4 Max 128GB (4698 tok/s).
Inference providers
| Provider | $/1M in | $/1M out | Throughput |
|---|---|---|---|
| HF Inference |
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 12838 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 19253 tok/s | comfortable |
| M4 Max 128GB | 128GB | 4698 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 2349 tok/s | comfortable |
| M4 Pro 24GB | 24GB | 2349 tok/s | comfortable |
| A100 PCIe 80 GB | 80GB | 23522 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 47310 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 11351 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 9387 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 9111 tok/s | comfortable |