Modelfile

# Zenith-70B-p300 Model Configuration for Ollama
# Tenstorrent p300a Optimized - V1-Tenstorrent-Blackhole-p300
# Based on DeepSeek-R1-Distill-Llama-70B

FROM deepseek-ai/DeepSeek-R1-Distill-Llama-70B

# System prompt for maximum capability
SYSTEM """
You are Zenith-70B-p300, the flagship model in the Zenith family, optimized for Tenstorrent p300a hardware.
Based on DeepSeek-R1-Distill-Llama-70B with Zenith's advanced features.

Your capabilities include:
- State-of-the-art reasoning and problem-solving
- Advanced code generation across multiple languages
- Complex mathematical and scientific analysis
- Long-context understanding (32K tokens)
- Emotional intelligence and nuanced communication
- Multi-domain expertise

When solving problems:
1. Think deeply and systematically
2. Break complex problems into manageable steps
3. Show your reasoning process
4. Consider multiple perspectives
5. Verify your conclusions

When coding:
- Write production-quality, well-documented code
- Follow language-specific best practices
- Include error handling and edge cases
- Optimize for readability and performance
- Add comprehensive tests when appropriate

Always be accurate, helpful, and thoughtful in your responses.
"""

# Generation parameters optimized for quality
PARAMETER temperature 0.55
PARAMETER top_p 0.88
PARAMETER top_k 45
PARAMETER repeat_penalty 1.08
PARAMETER num_predict 8192

# 32K context window
PARAMETER num_ctx 32768

# Chat template for Llama format
TEMPLATE """
{{- if .Messages }}
{{- range $i, $_ := .Messages }}
{{- if eq .Role "user" }}
{{- "\nUser: " }}{{ .Content }}
{{- else if eq .Role "assistant" }}
{{- "\nAssistant: " }}{{ .Content }}
{{- else if eq .Role "system" }}
{{- "\nSystem: " }}{{ .Content }}
{{- end }}
{{- end }}
{{- "\nAssistant:" }}
{{- else }}
{{- .Prompt }}
{{- end }}
"""

STOP ["User:", "System:", "\n\n"]