Superthoughts Lite V2 MOE Llama3.2 GGUF

🚀 Superthoughts Lite v2 MOE Llama3.2

A powerful and lite reasoning model for various tasks including chemistry, code, math, and conversations.

🚀 Quick Start

This is the GGUF version of Superthoughts Lite v2 MOE Llama3.2, with 3.91B parameters, 2 experts active, and 4 in total.

You can access different versions of the model:

• GGUF
• Full precision
• BF16

✨ Features

• Better Performance: This non - experimental version offers better accuracy at all tasks, better performance, and less looping while generating responses.
• Powerful Reasoning: It is a powerful, lite reasoning model trained with multiple experts, including chat, math, code, and science reasoning experts.
• Replacement for v1: It is a direct replacement of Pinkstack/Superthoughts - lite - v1, with much better code generation and text performance.

📚 Documentation

Model Training

We trained the model by first creating a base model for all the experts, which was fine - tuned using GRPO techniques with Unsloth on top of meta - llama/Llama - 3.2 - 1B - Instruct. After that, we trained each potential expert using SFT and then did GRPO again. There are 4 experts in total:

• Chat reasoning expert
• Math reasoning expert
• Code reasoning expert
• Science reasoning expert

System Prompt

You should use the following system prompt:

Thinking: enabled.

Follow this format strictly:
<think>
Write your step - by - step reasoning here.
Break down the problem into smaller parts.
Solve each part systematically.
Check your work and verify the answer makes sense.
</think>
[Your final answer after thinking].

Model Information

The model can generate up to 16,380 tokens and has a context size of 131072. It has been fine - tuned to generate thinking data in - between <think> xml tags. Note that it may still have some slight looping, but they are rare.

Limitations

• Safety Alignment: While some safety alignment was done, it was very minimal. Thus, the model can be uncensored at times.
• Hallucination: All large language models (LLMs), including this one, can hallucinate and output false information. Always double - check responses.
• Knowledge Dependency: The chat model may make things up unless you provide it with proper information.

License

By using this model, you agree to the [LLAMA 3.2 COMMUNITY LICENSE](https://huggingface.co/meta - llama/Llama - 3.2 - 1B/blob/main/LICENSE.txt).

GGUF Template

{{ if .Messages }}
{{- if or .System .Tools }}<|start_header_id|>system<|end_header_id|>
{{- if .System }}

{{ .System }}
{{- end }}
{{- if .Tools }}

You are a helpful assistant with tool calling capabilities. When you receive a tool call response, use the output to format an answer to the original use question.
{{- end }}
{{- end }}<|eot_id|>
{{- range $i, $_ := .Messages }}
{{- $last := eq (len (slice $.Messages $i)) 1 }}
{{- if eq .Role "user" }}<|start_header_id|>user<|end_header_id|>
{{- if and $.Tools $last }}

Given the following functions, please respond with a JSON for a function call with its proper arguments that best answers the given prompt.

Respond in the format {"name": function name, "parameters": dictionary of argument name and its value}. Do not use variables.

{{ $.Tools }}
{{- end }}

{{ .Content }}<|eot_id|>{{ if $last }}<|start_header_id|>assistant<|end_header_id|>

{{ end }}
{{- else if eq .Role "assistant" }}<|start_header_id|>assistant<|end_header_id|>
{{- if .ToolCalls }}

{{- range .ToolCalls }}{"name": "{{ .Function.Name }}", "parameters": {{ .Function.Arguments }}}{{ end }}
{{- else }}

{{ .Content }}{{ if not $last }}<|eot_id|>{{ end }}
{{- end }}
{{- else if eq .Role "tool" }}<|start_header_id|>ipython<|end_header_id|>

{{ .Content }}<|eot_id|>{{ if $last }}<|start_header_id|>assistant<|end_header_id|>

{{ end }}
{{- end }}
{{- end }}
{{- else }}
{{- if .System }}<|start_header_id|>system<|end_header_id|>

{{ .System }}<|eot_id|>{{ end }}{{ if .Prompt }}<|start_header_id|>user<|end_header_id|>

{{ .Prompt }}<|eot_id|>{{ end }}<|start_header_id|>assistant<|end_header_id|>

{{ end }}{{ .Response }}{{ if .Response }}<|eot_id|>{{ end }}

📄 License

The model is under the [LLAMA 3.2 COMMUNITY LICENSE](https://huggingface.co/meta - llama/Llama - 3.2 - 1B/blob/main/LICENSE.txt).

📦 Information Table

⚠️ Important Note

• The model has minimal safety alignment, so it can be uncensored at times.
• All large language models can hallucinate and output false information. Always double - check responses.
• The chat model may make things up if not provided with proper information.

💡 Usage Tip Make sure to use the provided system prompt for better results.