Qwen 2.5 72b RP Ink
A roleplay-focused LoRA finetune of Qwen 2.5 72b Instruct. Methodology and hyperparams inspired by SorcererLM and Slush.
Yet another model in the Ink series, following in the footsteps of the 32b one and the Nemo one
Testimonials
[Compared to the 32b] felt a noticeable increase in coherence
- ShotMisser64
Yeah ep2's great!! made me actually wanna write a reply by myself for the first time in a few days
- Maw
This is the best RP I've ever had
- 59smoke
this makes me want to get another 3090 to run 72b
- dysfunctional
Dataset
The worst mix of data you've ever seen. Like, seriously, you do not want to see the things that went into this model. It's bad.
"this is like washing down an adderall with a bottle of methylated rotgut" - inflatebot
Update: I have sent the (public datasets in the) data mix publicly already so here's that
Quants
imatrix GGUFs by bartowski
exl2s by sleep deprived
Recommended Settings
Chat template: ChatML
Recommended samplers (not the be-all-end-all, try some on your own!):
- Temp 0.83 / Top P 0.8 / Top A 0.3 / Rep Pen 1.03
- Your samplers can go here! :3
Hyperparams
General
- Epochs = 2
- LR = 6e-5
- LR Scheduler = Cosine
- Optimizer = Paged AdamW 8bit
- Effective batch size = 16
LoRA
- Rank = 16
- Alpha = 32
- Dropout = 0.25 (Inspiration: Slush)
Credits
Humongous thanks to the people who created and curated the original data
Big thanks to all Allura members, for testing and emotional support ilya /platonic
especially to inflatebot who made the model card's image :3
Another big thanks to all the members of the ArliAI and BeaverAI Discord servers for testing! All of the people featured in the testimonials are from there :3