Model comparison
nomic-embed-text-v1 vs jina-reranker-v1-tiny-en
At Q4_K_M, jina-reranker-v1-tiny-en is the smaller download — 33,257,536 bytes against 84,106,624. At long context the gap widens: jina-reranker-v1-tiny-en's KV cache at 32K is 6.0× smaller, which usually matters more than the difference in weights.
From the file· summed bytes, KV per layer
Side by side
| nomic-embed-text-v1 | jina-reranker-v1-tiny-en | |
|---|---|---|
| Parameters | 137M | 33M |
| Architecture | nomic-bert | jina-bert-v2 |
| Layers | 12 | 4 |
| Native context | 8,192 | 8,192 |
| Mixture of experts | no | no |
| Quantizations published | 14 | 12 |
| Smallest quantization | 0.05 GiB | 0.03 GiB |
| Q4_K_M | 0.08 GiB | 0.03 GiB |
| Licence | apache-2.0 | apache-2.0 |
KV cache by context
the term that decides long-context viability
| Context | nomic-embed-text-v1 | jina-reranker-v1-tiny-en | Ratio |
|---|---|---|---|
| 4,096 | 0.14 GiB | 0.02 GiB | 6.00× |
| 8,192 | 0.28 GiB | 0.05 GiB | 6.00× |
| 16,384 | 0.56 GiB | 0.09 GiB | 6.00× |
| 32,768 | 1.13 GiB | 0.19 GiB | 6.00× |
| 65,536 | 2.25 GiB | 0.38 GiB | 6.00× |
| 131,072 | 4.50 GiB | 0.75 GiB | 6.00× |