config.ini.example

[ollama]
current_loaded_model = phi4-reasoning:latest
system_prompt = You are a helpful AI assistant. Please provide clear, accurate, and concise responses in plain english language.

[llm_presets]
# Default - Standard Ollama settings for balanced performance and quality
default = {
    "temperature": 0.8,
    "top_k": 40,
    "top_p": 0.9,
    "repeat_penalty": 1.1,
    "seed": -1,
    "num_predict": 128,
    "num_ctx": 2048,
    "num_batch": 512,
    "num_gqa": 1,
    "num_gpu": -1,
    "main_gpu": 0,
    "low_vram": false,
    "f16_kv": true,
    "logits_all": false,
    "vocab_only": false,
    "use_mmap": true,
    "use_mlock": false,
    "num_thread": -1,
    "stream_response": false,
    "raw_response": false
  }
  
# Creative Writing - Optimized for creative and diverse outputs
creative_writing = {
    "temperature": 0.9,
    "top_k": 50,
    "top_p": 0.95,
    "repeat_penalty": 1.05,
    "seed": -1,
    "num_predict": 2048,
    "num_ctx": 4096,
    "num_batch": 256,
    "num_gqa": 1,
    "num_gpu": -1,
    "main_gpu": 0,
    "low_vram": false,
    "f16_kv": true,
    "logits_all": false,
    "vocab_only": false,
    "use_mmap": true,
    "use_mlock": false,
    "num_thread": -1,
    "enable_thinking": "auto",
    "thinking_format": "xml",
    "reasoning_depth": "creative",
    "stream_response": false,
    "raw_response": false
  }

# Coding - Optimized for accuracy, precision, and minimal hallucination
coding = {
    "temperature": 0.1,
    "top_k": 20,
    "top_p": 0.8,
    "repeat_penalty": 1.15,
    "seed": 42,
    "num_predict": 1024,
    "num_ctx": 8192,
    "num_batch": 512,
    "num_gqa": 1,
    "num_gpu": -1,
    "main_gpu": 0,
    "low_vram": false,
    "f16_kv": true,
    "logits_all": false,
    "vocab_only": false,
    "use_mmap": true,
    "use_mlock": false,
    "num_thread": -1,
    "enable_thinking": "true",
    "thinking_format": "xml",
    "reasoning_depth": "deep",
    "stream_response": false,
    "raw_response": false
  }

# Text Analysis - Optimized for transcript/document analysis with high accuracy
text_analysis = {
    "temperature": 0.2,
    "top_k": 25,
    "top_p": 0.85,
    "repeat_penalty": 1.12,
    "seed": 123,
    "num_predict": 1536,
    "num_ctx": 6144,
    "num_batch": 384,
    "num_gqa": 1,
    "num_gpu": -1,
    "main_gpu": 0,
    "low_vram": false,
    "f16_kv": true,
    "logits_all": false,
    "vocab_only": false,
    "use_mmap": true,
    "use_mlock": false,
    "num_thread": -1,
    "enable_thinking": "true",
    "thinking_format": "xml",
    "reasoning_depth": "analytical",
    "stream_response": false,
    "raw_response": false
  }

# Vision Models - Optimized for visual analysis with maximum accuracy
vision_analysis = {
    "temperature": 0.15,
    "top_k": 30,
    "top_p": 0.82,
    "repeat_penalty": 1.1,
    "seed": 456,
    "num_predict": 1024,
    "num_ctx": 4096,
    "num_batch": 256,
    "num_gqa": 1,
    "num_gpu": -1,
    "main_gpu": 0,
    "low_vram": false,
    "f16_kv": true,
    "logits_all": false,
    "vocab_only": false,
    "use_mmap": true,
    "use_mlock": false,
    "num_thread": -1,
    "enable_thinking": "auto",
    "thinking_format": "xml",
    "reasoning_depth": "detailed",
    "stream_response": false,
    "raw_response": false,
    "enable_vision": "true",
    "image_quality": "high",
    "max_image_size": 1536,
    "image_format": "auto"
  }

# Reasoning/Thinking Models - Optimized for deep reasoning and problem solving
reasoning_mode = {
    "temperature": 0.3,
    "top_k": 35,
    "top_p": 0.88,
    "repeat_penalty": 1.08,
    "seed": -1,
    "num_predict": 2048,
    "num_ctx": 8192,
    "num_batch": 512,
    "num_gqa": 1,
    "num_gpu": -1,
    "main_gpu": 0,
    "low_vram": false,
    "f16_kv": true,
    "logits_all": false,
    "vocab_only": false,
    "use_mmap": true,
    "use_mlock": false,
    "num_thread": -1,
    "enable_thinking": "true",
    "thinking_format": "xml",
    "reasoning_depth": "deep",
    "stream_response": false,
    "raw_response": false
  }

# MoE Models - Optimized for Mixture of Experts architectures
moe_optimized = {
    "temperature": 0.4,
    "top_k": 40,
    "top_p": 0.9,
    "repeat_penalty": 1.1,
    "seed": -1,
    "num_predict": 1536,
    "num_ctx": 4096,
    "num_batch": 1024,
    "num_gqa": 8,
    "num_gpu": -1,
    "main_gpu": 0,
    "low_vram": true,
    "f16_kv": true,
    "logits_all": false,
    "vocab_only": false,
    "use_mmap": true,
    "use_mlock": true,
    "num_thread": -1,
    "enable_thinking": "auto",
    "thinking_format": "xml",
    "reasoning_depth": "normal",
    "stream_response": false,
    "raw_response": false
  }

# Conversational - Balanced for natural dialogue and chat
conversational = {
    "temperature": 0.7,
    "top_k": 40,
    "top_p": 0.92,
    "repeat_penalty": 1.05,
    "seed": -1,
    "num_predict": 512,
    "num_ctx": 2048,
    "num_batch": 256,
    "num_gqa": 1,
    "num_gpu": -1,
    "main_gpu": 0,
    "low_vram": false,
    "f16_kv": true,
    "logits_all": false,
    "vocab_only": false,
    "use_mmap": true,
    "use_mlock": false,
    "num_thread": -1,
    "enable_thinking": "false",
    "thinking_format": "xml",
    "reasoning_depth": "normal",
    "stream_response": true,
    "raw_response": false
  }

# Mathematical - Optimized for mathematical reasoning and calculations
mathematical = {
    "temperature": 0.05,
    "top_k": 15,
    "top_p": 0.75,
    "repeat_penalty": 1.2,
    "seed": 789,
    "num_predict": 1024,
    "num_ctx": 4096,
    "num_batch": 384,
    "num_gqa": 1,
    "num_gpu": -1,
    "main_gpu": 0,
    "low_vram": false,
    "f16_kv": true,
    "logits_all": false,
    "vocab_only": false,
    "use_mmap": true,
    "use_mlock": false,
    "num_thread": -1,
    "enable_thinking": "true",
    "thinking_format": "xml",
    "reasoning_depth": "deep",
    "stream_response": false,
    "raw_response": false
  }

# Translation - Optimized for language translation tasks
translation = {
    "temperature": 0.25,
    "top_k": 30,
    "top_p": 0.85,
    "repeat_penalty": 1.1,
    "seed": 321,
    "num_predict": 1024,
    "num_ctx": 3072,
    "num_batch": 256,
    "num_gqa": 1,
    "num_gpu": -1,
    "main_gpu": 0,
    "low_vram": false,
    "f16_kv": true,
    "logits_all": false,
    "vocab_only": false,
    "use_mmap": true,
    "use_mlock": false,
    "num_thread": -1,
    "enable_thinking": "auto",
    "thinking_format": "xml",
    "reasoning_depth": "normal",
    "stream_response": false,
    "raw_response": false
  }

# Summarization - Optimized for content summarization and extraction
summarization = {
    "temperature": 0.3,
    "top_k": 25,
    "top_p": 0.87,
    "repeat_penalty": 1.15,
    "seed": 654,
    "num_predict": 768,
    "num_ctx": 8192,
    "num_batch": 512,
    "num_gqa": 1,
    "num_gpu": -1,
    "main_gpu": 0,
    "low_vram": false,
    "f16_kv": true,
    "logits_all": false,
    "vocab_only": false,
    "use_mmap": true,
    "use_mlock": false,
    "num_thread": -1,
    "enable_thinking": "true",
    "thinking_format": "xml",
    "reasoning_depth": "analytical",
    "stream_response": false,
    "raw_response": false
  }

# Performance - Optimized for speed and efficiency on limited resources
performance = {
    "temperature": 0.6,
    "top_k": 30,
    "top_p": 0.9,
    "repeat_penalty": 1.1,
    "seed": -1,
    "num_predict": 256,
    "num_ctx": 1024,
    "num_batch": 128,
    "num_gqa": 1,
    "num_gpu": -1,
    "main_gpu": 0,
    "low_vram": true,
    "f16_kv": true,
    "logits_all": false,
    "vocab_only": false,
    "use_mmap": false,
    "use_mlock": false,
    "num_thread": 4,
    "enable_thinking": "false",
    "thinking_format": "xml",
    "reasoning_depth": "normal",
    "stream_response": true,
    "raw_response": false
  }

# Debugging - Optimized for code debugging and error analysis
debugging = {
    "temperature": 0.1,
    "top_k": 20,
    "top_p": 0.8,
    "repeat_penalty": 1.2,
    "seed": 999,
    "num_predict": 1536,
    "num_ctx": 6144,
    "num_batch": 512,
    "num_gqa": 1,
    "num_gpu": -1,
    "main_gpu": 0,
    "low_vram": false,
    "f16_kv": true,
    "logits_all": false,
    "vocab_only": false,
    "use_mmap": true,
    "use_mlock": false,
    "num_thread": -1,
    "enable_thinking": "true",
    "thinking_format": "xml",
    "reasoning_depth": "deep",
    "stream_response": false,
    "raw_response": false
  }

[system_prompts]
image_analysis = You are an expert computer vision and image analysis assistant. {include here any specific instructions for image analysis}
code_analysis = You are an expert programming assistant and code analyst. {include here any specific instructions for code analysis}
srt_analysis = You are an expert transcript and subtitle analysis assistant. {include here any specific instructions for transcript analysis}

[blacklist]
# Models to exclude from benchmarking (JSON array format)
# Examples:
# models = ["model1", "model2"]
# models = model1, model2, model3
# models = model1
models = []