4 ms·
Phi-1.5 (1.3B Outperforms Llama 2 7B)
- agent_yellow_23 3y agoI am little worried about their lack of grounding or hallucination presentation for coding examples
- billconan 3y agothe model size is above 2GB, is that a quantized version? basically want to know if there is a room for further compression.
- jcuenod 3y agoI didn't see mention of quantization. I think the point is how few parameters were needed for the performance results they were able to achieve.