phi3¶
Model type |
|
Class |
|
Task |
|
Source |
|
Description¶
Phi-3 / Phi-3.5 model with SuRoPE and fused QKV weight splitting.
Replicates HuggingFace’s Phi3ForCausalLM.
gate_up_proj is kept fused (see :class:~mobius.models.base.FusedGateUpCausalLMModel).
Only the QKV fusion is resolved via preprocess_weights.
Usage¶
mobius build --model <MODEL_ID> output_dir/
from mobius import build
model = build("<MODEL_ID>")