Model comparison
whisper-large-v2 vs nemotron-3.5-asr-streaming-0.6b
At Q4_K_M, nemotron-3.5-asr-streaming-0.6b is the smaller download — 495,831,520 bytes against 996,526,080.
From the file· summed bytes, KV per layer
Side by side
| whisper-large-v2 | nemotron-3.5-asr-streaming-0.6b | |
|---|---|---|
| Parameters | 1.5B | 638M |
| Architecture | whisper | parakeet |
| Layers | 32 | — |
| Native context | — | — |
| Mixture of experts | no | no |
| Quantizations published | 6 | 8 |
| Smallest quantization | 0.93 GiB | 0.38 GiB |
| Q4_K_M | 0.93 GiB | 0.46 GiB |
| Licence | apache-2.0 | other |
KV cache by context
the term that decides long-context viability
| Context | whisper-large-v2 | nemotron-3.5-asr-streaming-0.6b | Ratio |
|---|---|---|---|
| 4,096 | — | — | — |
| 8,192 | — | — | — |
| 16,384 | — | — | — |
| 32,768 | — | — | — |
| 65,536 | — | — | — |
| 131,072 | — | — | — |