Delta-Vector/Francois-PE-V2

Francois-PE

This is the base model for Francois-PE-Huali, I'd reccomend using that instead of this, The model is extremely underfit and via KTO - was fixed and wayyyyyyyyyyy more coherent

List of datasets:

datasets:
- PocketDoc/Dans-Personamaxx-VN
- NewEden/LIMARP-Complexity
- NewEden/PIPPA-Mega-Filtered
- NewEden/OpenCAI-ShareGPT
- NewEden/Creative_Writing-Complexity
- NewEden/Light-Novels-Roleplay-Logs-Books-Oh-My-duplicate-turns-removed
- PocketDoc/Dans-Failuremaxx-Adventure-3
- NewEden/Books-V2-ShareGPT
- NewEden/Deepseek-V3-RP-Filtered
- NewEden/BlueSky-10K-Complexity
- NewEden/Final-Alpindale-LNs-ShareGPT
- NewEden/DeepseekRP-Filtered
- NewEden/RP-logs-V2-Experimental
- anthracite-org/kalo_opus_misc_240827
- anthracite-org/kalo_misc_part2
- NewEden/vanilla-backrooms-claude-sharegpt
- NewEden/Storium-Prefixed-Clean
- NewEden/KTO-IF-Dans
- NewEden/KTO-Instruct-Mix
- NewEden/Opus-accepted-hermes-rejected-shuffled

KTO wandb: https://wandb.ai/new-eden/Francois-V2/runs/f2qejmu0?nw=nwuserdeltavector SFT wandb: https://wandb.ai/new-eden/Francois/runs/sn3utrs1?nw=nwuserdeltavector