
Smarter-Phi-3 By Utilizing Intuition on the Signal To Noise Ratio To Optimize Depth Up Scaling To Use The Most Valuable Layers
af85264
verified
slices: | |
- sources: | |
- model: TroyDoesAI/Phi-3-Context-Obedient-RAG | |
layer_range: [0, 8] | |
- sources: | |
- model: TroyDoesAI/Phi-3-Context-Obedient-RAG | |
layer_range: [4, 12] | |
- sources: | |
- model: TroyDoesAI/Phi-3-Context-Obedient-RAG | |
layer_range: [8, 16] | |
- sources: | |
- model: TroyDoesAI/Phi-3-Context-Obedient-RAG | |
layer_range: [12, 20] | |
- sources: | |
- model: TroyDoesAI/Phi-3-Context-Obedient-RAG | |
layer_range: [16, 24] | |
- sources: | |
- model: TroyDoesAI/Phi-3-Context-Obedient-RAG | |
layer_range: [20, 28] | |
- sources: | |
- model: TroyDoesAI/Phi-3-Context-Obedient-RAG | |
layer_range: [24, 32] | |
merge_method: passthrough | |
dtype: bfloat16 |