docs: sync Modelfile with updated benchmark numbers

Modelfile

@@ -0,0 +1,28 @@
+# Ollama Modelfile for Simplicio 27B
+# Optimized for: Atomic Search/Replace Diffs & Simplicio-Loop
+# Quantization recommended: Q4_K_M (~16.8 GB) or Q8_0 (~28.5 GB)
+
+FROM ./simplicio-27b-Q4_K_M.gguf
+
+# Generation Parameters
+PARAMETER temperature 0.2
+PARAMETER top_p 0.95
+PARAMETER top_k 40
+PARAMETER repeat_penalty 1.1
+PARAMETER num_ctx 32768
+
+# Deterministic Stop Tokens
+PARAMETER stop "<|im_end|>"
+PARAMETER stop "</deliver>"
+PARAMETER stop "<|endoftext|>"
+
+# ChatML Template with Simplicio Loop Tokens Support
+TEMPLATE """{{ if .System }}<|im_start|>system
+{{ .System }}<|im_end|>
+{{ end }}{{ if .Prompt }}<|im_start|>user
+{{ .Prompt }}<|im_end|>
+{{ end }}<|im_start|>assistant
+{{ .Response }}<|im_end|>"""
+
+# Official System Prompt
+SYSTEM """Voce e o Simplicio 27B, treinado para executar tarefas de desenvolvimento seguindo rigorosamente os 50 pontos do Simplicio-Loop: Orientacao (<orient>), Planejamento (<plan>), Edicao Cirurgica por Diff (<patch>), Validacao (<validate>) e Entrega Verificada (<deliver>) sem alucinacao e com zero desperdicio de tokens."""