#--DUAL MODEL MERGE SETUP--- | |
# TRY AGAIN, BUT COPY OVER THE CORRECT .JSON FILES FROM | |
# HARMONIC HARLEQUIN SO IT CAN GGUF.. maybe.. | |
# The models we are going to use. | |
const_tag: &BASE_MODEL Undi95/PsyMedRP-v1-20B | |
const_tag: &MODEL1 Elfrino/XwinXtended-20B # Will this guy cause me headaches? | |
const_tag: &MODEL2 Undi95/PsyMedRP-v1-20B | |
# The amount to scale the contribution to the residual stream (to hopefully reduce overshoot). | |
const_tag: &RESIDUAL_SCALE_FACTOR 0.71 # back to 0.7 | |
model1-filter-env: | |
parameters: | |
scale: | |
- filter: down_proj | |
value: | |
- value: 1.0 | |
model2-filter-env: | |
parameters: | |
scale: | |
- filter: down_proj | |
value: | |
- value: 1.0 | |
slices: | |
# The first 10 layers are not duplicated. | |
- sources: | |
- model: | |
layer_range: [0, 10] | |
- sources: | |
- model: | |
layer_range: [10, 11] | |
<<: | |
- sources: | |
- model: | |
layer_range: [10, 11] | |
<<: | |
- sources: | |
- model: | |
layer_range: [11, 12] | |
<<: | |
- sources: | |
- model: | |
layer_range: [11, 12] | |
<<: | |
- sources: | |
- model: | |
layer_range: [12, 13] | |
<<: | |
- sources: | |
- model: | |
layer_range: [12, 13] | |
<<: | |
- sources: | |
- model: | |
layer_range: [13, 14] | |
<<: | |
- sources: | |
- model: | |
layer_range: [13, 14] | |
<<: | |
- sources: | |
- model: | |
layer_range: [14, 15] | |
<<: | |
- sources: | |
- model: | |
layer_range: [14, 15] | |
<<: | |
- sources: | |
- model: | |
layer_range: [15, 16] | |
<<: | |
- sources: | |
- model: | |
layer_range: [15, 16] | |
<<: | |
- sources: | |
- model: | |
layer_range: [16, 17] | |
<<: | |
- sources: | |
- model: | |
layer_range: [16, 17] | |
<<: | |
- sources: | |
- model: | |
layer_range: [17, 18] | |
<<: | |
- sources: | |
- model: | |
layer_range: [17, 18] | |
<<: | |
- sources: | |
- model: | |
layer_range: [18, 19] | |
<<: | |
- sources: | |
- model: | |
layer_range: [18, 19] | |
<<: | |
- sources: | |
- model: | |
layer_range: [19, 20] | |
<<: | |
- sources: | |
- model: | |
layer_range: [19, 20] | |
<<: | |
- sources: | |
- model: | |
layer_range: [20, 21] | |
<<: | |
- sources: | |
- model: | |
layer_range: [20, 21] | |
<<: | |
- sources: | |
- model: | |
layer_range: [21, 22] | |
<<: | |
- sources: | |
- model: | |
layer_range: [21, 22] | |
<<: | |
- sources: | |
- model: | |
layer_range: [22, 23] | |
<<: | |
- sources: | |
- model: | |
layer_range: [22, 23] | |
<<: | |
- sources: | |
- model: | |
layer_range: [23, 24] | |
<<: | |
- sources: | |
- model: | |
layer_range: [23, 24] | |
<<: | |
- sources: | |
- model: | |
layer_range: [24, 25] | |
<<: | |
- sources: | |
- model: | |
layer_range: [24, 25] | |
<<: | |
- sources: | |
- model: | |
layer_range: [25, 26] | |
<<: | |
- sources: | |
- model: | |
layer_range: [25, 26] | |
<<: | |
- sources: | |
- model: | |
layer_range: [26, 27] | |
<<: | |
- sources: | |
- model: | |
layer_range: [26, 27] | |
<<: | |
- sources: | |
- model: | |
layer_range: [27, 28] | |
<<: | |
- sources: | |
- model: | |
layer_range: [27, 28] | |
<<: | |
- sources: | |
- model: | |
layer_range: [28, 29] | |
<<: | |
- sources: | |
- model: | |
layer_range: [28, 29] | |
<<: | |
- sources: | |
- model: | |
layer_range: [29, 30] | |
<<: | |
- sources: | |
- model: | |
layer_range: [29, 30] | |
<<: | |
- sources: | |
- model: | |
layer_range: [30, 31] | |
<<: | |
- sources: | |
- model: | |
layer_range: [30, 31] | |
<<: | |
- sources: | |
- model: | |
layer_range: [31, 32] | |
<<: | |
- sources: | |
- model: | |
layer_range: [31, 32] | |
<<: | |
- sources: | |
- model: | |
layer_range: [32, 33] | |
<<: | |
- sources: | |
- model: | |
layer_range: [32, 33] | |
<<: | |
- sources: | |
- model: | |
layer_range: [33, 34] | |
<<: | |
- sources: | |
- model: | |
layer_range: [33, 34] | |
<<: | |
- sources: | |
- model: | |
layer_range: [34, 35] | |
<<: | |
- sources: | |
- model: | |
layer_range: [34, 35] | |
<<: | |
- sources: | |
- model: | |
layer_range: [35, 36] | |
<<: | |
- sources: | |
- model: | |
layer_range: [35, 36] | |
<<: | |
- sources: | |
- model: | |
layer_range: [36, 37] | |
<<: | |
- sources: | |
- model: | |
layer_range: [36, 37] | |
<<: | |
- sources: | |
- model: | |
layer_range: [37, 38] | |
<<: | |
- sources: | |
- model: | |
layer_range: [37, 38] | |
<<: | |
- sources: | |
- model: | |
layer_range: [38, 39] | |
<<: | |
- sources: | |
- model: | |
layer_range: [38, 39] | |
<<: | |
- sources: | |
- model: | |
layer_range: [39, 40] | |
<<: | |
- sources: | |
- model: | |
layer_range: [39, 40] | |
<<: | |
- sources: | |
- model: | |
layer_range: [40, 41] | |
<<: | |
- sources: | |
- model: | |
layer_range: [40, 41] | |
<<: | |
- sources: | |
- model: | |
layer_range: [41, 42] | |
<<: | |
- sources: | |
- model: | |
layer_range: [41, 42] | |
<<: | |
- sources: | |
- model: | |
layer_range: [42, 43] | |
<<: | |
- sources: | |
- model: | |
layer_range: [42, 43] | |
<<: | |
- sources: | |
- model: | |
layer_range: [43, 44] | |
<<: | |
- sources: | |
- model: | |
layer_range: [43, 44] | |
<<: | |
- sources: | |
- model: | |
layer_range: [44, 45] | |
<<: | |
- sources: | |
- model: | |
layer_range: [44, 45] | |
<<: | |
- sources: | |
- model: | |
layer_range: [45, 46] | |
<<: | |
- sources: | |
- model: | |
layer_range: [45, 46] | |
<<: | |
- sources: | |
- model: | |
layer_range: [46, 47] | |
<<: | |
- sources: | |
- model: | |
layer_range: [46, 47] | |
<<: | |
- sources: | |
- model: | |
layer_range: [47, 48] | |
<<: | |
- sources: | |
- model: | |
layer_range: [47, 48] | |
<<: | |
- sources: | |
- model: | |
layer_range: [48, 49] | |
<<: | |
- sources: | |
- model: | |
layer_range: [48, 49] | |
<<: | |
- sources: | |
- model: | |
layer_range: [49, 50] | |
<<: | |
- sources: | |
- model: | |
layer_range: [49, 50] | |
<<: | |
- sources: | |
- model: | |
layer_range: [50, 51] | |
<<: | |
- sources: | |
- model: | |
layer_range: [50, 51] | |
<<: | |
- sources: | |
- model: | |
layer_range: [51, 52] | |
<<: | |
- sources: | |
- model: | |
layer_range: [51, 52] | |
<<: | |
# The last 10 layers are not duplicated. | |
- sources: | |
- model: | |
layer_range: [52, 62] | |
merge_method: passthrough | |
dtype: bfloat16 |