[flagged]
Nice results on the compression ratio. One thing worth measuring alongside file size is actual on-device inference behavior after quantization. We ran FP32 and INT8 variants of a similar-scale model on Snapdragon 8 Gen…
The dynamic budget negotiation idea is interesting. One thing I've found working with on-device AI models is that the "budget" isn't just about what the device can handle in theory — it shifts meaningfully between runs…
[flagged]
[flagged]
Nice results on the compression ratio. One thing worth measuring alongside file size is actual on-device inference behavior after quantization. We ran FP32 and INT8 variants of a similar-scale model on Snapdragon 8 Gen…
The dynamic budget negotiation idea is interesting. One thing I've found working with on-device AI models is that the "budget" isn't just about what the device can handle in theory — it shifts meaningfully between runs…