All credits belong to the author.
It stands for Importance Matrix, a technique used to improve the quality of quantized models.
[1]
The Imatrix is calculated based on calibration data, and it helps determine the importance of different model activations during the quantization process. The idea is to preserve the most important information during quantization, which can help reduce the loss of model performance and lead to better performance, especially when the calibration data is diverse.
[2]
For --imatrix data, imatrix-Eris_7B-F16.dat was used.
Using llama.cpp-bb2308:
Base⇢ GGUF(F16)⇢ Imatrix-Data(F16)⇢ GGUF(Imatrix-Quants)
The new IQ3_S quant-option has shown to be better than the old Q3_K_S, so I added that instead of the later. Only supported in koboldcpp-1.59.1 or higher.
If you want any specific quantization to be added, feel free to ask.
Notes: Model should be excellent for both RP/Chat related tasks. Seems to be working in both Alpaca/Chatml.
Collaborative effort from both @Jeiku and @Nitral involving what we currently felt were our best individual projects.
We hope you enjoy! - The Chaotic Neutrals.
The following models were included in the merge:
The following YAML configuration was used to produce this model:
slices:
- sources:
- model: Test157t/Prima-LelantaclesV6-7b
layer_range: [0, 32]
- model: ChaoticNeutrals/Prodigy_7B
layer_range: [0, 32]
merge_method: slerp
base_model: Test157t/Prima-LelantaclesV6-7b
parameters:
t:
- filter: self_attn
value: [0, 0.5, 0.3, 0.7, 1]
- filter: mlp
value: [1, 0.5, 0.7, 0.3, 0]
- value: 0.5
dtype: bfloat16