metadata

tags:
  - iMat
  - GGUF
  - merge

  e88 88e                               d8     
 d888 888b  8888 8888  ,"Y88b 888 8e   d88     
C8888 8888D 8888 8888 "8" 888 888 88b d88888   
 Y888 888P  Y888 888P ,ee 888 888 888  888     
  "88 88"    "88 88"  "88 888 888 888  888     
      b                                        
      8b,                                      
 
  e88'Y88                  d8           888    
 d888  'Y  ,"Y88b 888,8,  d88    ,e e,  888    
C8888     "8" 888 888 "  d88888 d88 88b 888    
 Y888  ,d ,ee 888 888     888   888   , 888    
  "88,d88 "88 888 888     888    "YeeP" 888    
                                               
PROUDLY PRESENTS

0x01-8x7b-iMat-GGUF

Quantized from fp16 with love.

Quantizations made possible using .imatrix file from this repo (special thanks to ikawrakow again)

For a brief rundown of iMatrix quant performance please see this PR

All quants are verified working prior to uploading to repo for your safety and convenience.

Please note importance matrix quantizations are a work in progress, IQ3 and above is recommended for best results.

Tip: Pick a size that can fit in your GPU while still allowing some room for context for best speed. You may need to pad this further depending on if you are running image gen or TTS as well.

Original model card can be found here and below. Check there for optimal settings.

0x01-8x7B-hf

here we go again. multi-step merge, various models involved at various ratios with various methods.

this thing came to me in a fever dream when I was hung over, but after slightly tweaking the recipe it turned out surprisingly decent. using with the settings included.

Update:

The following settings have proved to work good too:

Context: https://files.catbox.moe/q91rca.json
Instruct: https://files.catbox.moe/2w8ja2.json
Textgen: https://files.catbox.moe/s25rad.json

Constituent parts

# primordial_slop_a:
  - model: mistralai/Mixtral-8x7B-v0.1+retrieval-bar/Mixtral-8x7B-v0.1_case-briefs
  - model: mistralai/Mixtral-8x7B-v0.1+SeanWu25/Mixtral_8x7b_Medicine
  - model: mistralai/Mixtral-8x7B-v0.1+SeanWu25/Mixtral_8x7b_WuKurtz
  - model: mistralai/Mixtral-8x7B-v0.1+Epiculous/crunchy-onion-lora
  - model: mistralai/Mixtral-8x7B-v0.1+maxkretchmer/gc-mixtral
# primordial_slop_b:
  - model: Envoid/Mixtral-Instruct-ITR-8x7B
  - model: crestf411/daybreak-mixtral-8x7b-v1.0-hf
  - model: NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO
  - model: orangetin/OpenHermes-Mixtral-8x7B
  - model: mistralai/Mixtral-8x7B-Instruct-v0.1+idegroup/PhyAssistant
  - model: ycros/crunchy-onion-nx
  - model: jondurbin/bagel-dpo-8x7b-v0.2
  - model: amoldwalunj/Mixtral-8x7B-Instruct-v0.1-legal_finetune_mixtral_32k
# primordial_slop_c: a+b
# primordial_slop_d:
  - model: Sao10K/Sensualize-Mixtral-bf16
  - model: Envoid/Mixtral-Instruct-ITR-DADA-8x7B

mergekit

This is a merge of pre-trained language models created using mergekit.

Merge Details

Merge Method

This model was merged using the SLERP merge method.

Models Merged

The following models were included in the merge:

./primordial_slop_d
./primordial_slop_c

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: ./primordial_slop_c
  - model: ./primordial_slop_d
merge_method: slerp
base_model: ./primordial_slop_c
parameters:
  t:
    - value: 0.33
dtype: float16