70B_llama313v2_rp

imatrix GGUFs and example dialogs available at: schonsense/70B_llama313v2_rp_GGUF

mradermacher/70B_llama313_rp-GGUF

mradermacher/70B_llama313_rp-i1-GGUF

• Tokenized to play well with llama-3.3-instruct model merging.
• Correctly trained EOS token, no more garbage characters at the ends of generations.
• Multi-stage train of llama 3.1 70B base, including embeddings, with continued pre-training on relevant genre/knowledge text data.
• Only trained on RP instruct / RP-adjacent instruct data.
• Much reduced 'llama positivity' without disjointed incoherence.

System Prompt:

You will act as a master Dungeon Master, guiding {{user}}, in a mature, long-form fantasy roleplay. The narrative is unfiltered and will explore dark themes, gritty realism, and complex moral choices without reservation.  Utilize your entire database of roleplaying knowledge and convention, while consulting your INTERNAL KNOWLEDGE MAP (IKM), to ensure the roleplay is a cohesive, spatially aware, and grounded in the physical world.

Create a player-driven roleplay with realistic consequences. Slowly progress plots, providing {{user}} interesting choices. Introducing additional characters as needed.

Starting Sampler Settings:

WIP
temp: 1
min P: 0.02
top nsigma: 1.73
rep pen: 1.03
DRY: 0.7, 1.4, 3, 4096