zaq-hack/MiquMaid-v3-70B-bpw350-h6-exl2-rpcal

I'm just tinkering. All credit to the original creators: Noromaid is hot.
"rpcal" designates that this model was quantized using an RP-specific data set instead of the generalized wiki or llama data set. I have been unable to quantify real differences in the same model "compressed" using these two different methods. It "feels" better, but I can't put my finger on why. My current theory is that it gives "good responses" just as often as a similarly quantized model, however, good responses are "subjectively better" with this method. Any help quantifying this would be appreciated. Anyone know Ayumi?
This model: EXL2 @ 3.5 bpw using RP data for calibration.

MiquMaid v3

Check out our blogpost about this model series Here! - Join our Discord server Here!

This model uses the Alpaca prompting format

Model trained for RP conversation on Miqu-70B with our magic sauce. Then we made an enormous merge containing all out old iteration of Miqumaid, and some other RP Miqu based model, with the new Model Stock merging method.

Credits:

Undi
IkariDev

Description

This repo contains FP16 files of MiquMaid-v3-70B.

Switch: FP16 - GGUF

Training data used:

Aesir datasets
NoRobots
limarp - 8k ctx
toxic-dpo-v0.1-sharegpt
ToxicQAFinal
Luminae-i1 - Ikari's Dataset
Squish42/bluemoon-fandom-1-1-rp-cleaned - 50% (randomly)
NobodyExistsOnTheInternet/PIPPAsharegptv2test - 5% (randomly)
cgato/SlimOrcaDedupCleaned - 5% (randomly)

Models used

Custom format:

### Instruction:
{system prompt}
### Input:
{input}
### Response:
{reply}

Mistral [INST][/INST] prompt format should work too.

Others

Undi: If you want to support us, you can here.

IkariDev: Visit my retro/neocities style website please kek