# Plumb-4B (v5, 4B, Q4_K_M), converted from crh225/plumb-4b with llama.cpp.
#
# A decision model, not a chat model, continued from JevK5 v0.2, so it uses
# JevK5's exact prompt: an ordinary chat request carrying only the question's
# JSON (evidence, criterion, lettered options) renders the right prompt inside
# Ollama, the same way the JevK5 Modelfile does.
#
# Callers read the answer letter's log-probabilities from one generated token
# and divide by Plumb-4B's calibration temperature T = 2.07.
FROM ./plumb-4b-v5-Q4_K_M.gguf
SYSTEM """Apply the supplied criterion to the supplied evidence. Choose exactly one listed option. Respond with only its uppercase letter, with no explanation or reasoning."""
TEMPLATE """<|im_start|>system
{{ .System }}<|im_end|>
<|im_start|>user
{{ .Prompt }}<|im_end|>
<|im_start|>assistant
<think>

</think>

"""
PARAMETER temperature 0
PARAMETER num_ctx 4096