Model comparison
whisper-base vs moonshine-streaming-tiny
These two publish different quantization sets; the table below has the exact sizes.
From the file· summed bytes, KV per layer
Side by side
| whisper-base | moonshine-streaming-tiny | |
|---|---|---|
| Parameters | 73M | 44M |
| Architecture | whisper | moonshine_streaming |
| Layers | 6 | 6 |
| Native context | — | 4,096 |
| Mixture of experts | no | no |
| Quantizations published | 19 | 4 |
| Smallest quantization | 0.03 GiB | 0.03 GiB |
| Q4_K_M | 0.05 GiB | — |
| Licence | apache-2.0 | mit |
KV cache by context
the term that decides long-context viability
| Context | whisper-base | moonshine-streaming-tiny | Ratio |
|---|---|---|---|
| 4,096 | — | 0.03 GiB | — |
| 8,192 | — | 0.06 GiB | — |
| 16,384 | — | 0.12 GiB | — |
| 32,768 | — | 0.23 GiB | — |
| 65,536 | — | 0.47 GiB | — |
| 131,072 | — | 0.94 GiB | — |