Model comparison
Fun-ASR-Nano-2512 vs parakeet-tdt-0.6b-v3
At Q4_K_M, parakeet-tdt-0.6b-v3 is the smaller download — 485,425,504 bytes against 556,974,848.
From the file· summed bytes, KV per layer
Side by side
| Fun-ASR-Nano-2512 | parakeet-tdt-0.6b-v3 | |
|---|---|---|
| Parameters | 830M | 627M |
| Architecture | funasr_nano | parakeet |
| Layers | — | — |
| Native context | — | — |
| Mixture of experts | no | no |
| Quantizations published | 9 | 10 |
| Smallest quantization | 0.52 GiB | 0.39 GiB |
| Q4_K_M | 0.52 GiB | 0.45 GiB |
| Licence | other | cc-by-4.0 |
KV cache by context
the term that decides long-context viability
| Context | Fun-ASR-Nano-2512 | parakeet-tdt-0.6b-v3 | Ratio |
|---|---|---|---|
| 4,096 | — | — | — |
| 8,192 | — | — | — |
| 16,384 | — | — | — |
| 32,768 | — | — | — |
| 65,536 | — | — | — |
| 131,072 | — | — | — |