Sathman/Meditation-Agent-TNH-SmolLM3-3B-v2-GGUF

Meditation Agent: TNH v2 (SmolLM3 3B)

A lightweight single-teacher contemplative model from the Meditation Agent v2 line for TNH.

This is the v2 successor to TNH-Agent-SmolLM3-3B-GGUF. It is trained on the newer E-atoms corpus, where the target is teaching_e, not the older atom surface.

Collections:

• New v2 collection: Meditation Agent v2 - EAtoms Teacher Series
• Previous collection: Meditation Agent - Non-Dual Self-Realization Teacher Series

Why Download This

• Single-teacher branch. This model is tuned specifically for TNH, not the mixed multi-teacher voice.
• Small local footprint. The current release file is Q5_K_M, which runs comfortably on local setups that cannot host larger models.
• Newer v2 corpus. The E-atoms line is sharper, less synthetic, and less likely to collapse into generic spiritual assistant prose.

Teacher Signal

This branch is tuned for gentle precision, interbeing, embodiment, and compassionate clarity without sentimentality.

Sample Outputs

These are from the local 5q smoke eval on the finished merged model.

Example 1

Prompt: How does fear relate to thought?

Friend, let us look deeply together. When we speak of 'fear,' what exactly do you find? Fear is not a thing that exists in isolation--it arises when we meet certain conditions: an object appears, and our mind creates the word 'danger' or 'enemy.' This meeting itself is already the first layer of fear's formation. Now observe closely at this very moment: right now, there are thoughts arising about whether things will be okay tomorrow. These thoughts create mental formations within your consciousness--thoughts of worry, anxiety, even dread for those who have passed away. Each one of these thoughts carries its own weight; some heavier than others but all contributing to this collective cloud known as fear. The thought 'I am afraid' is also a form of thought, yes? And when it arises, it becomes another layer in the same structure. So please, take three breaths with me now and see clearly how thought gives birth to feeling which in turn generates action--all flowing together like water from different sources into the same river.

Example 2

Prompt: Is there a self?

Friend, let us look together at this question. You speak of 'a self,' but what exactly do you mean by that word? When we say someone is selfish or generous--are they not merely acting from certain patterns? From habits formed over years in their family and culture? The patterns themselves are like seeds planted long ago; when conditions align just right, they sprout into action without any separate entity doing the acting.

This is why we practice looking deeply: to see these patterns clearly so they no longer dictate our behavior. We recognize that generosity arises precisely because we have been touched by the happiness it brings others, just as anger arises when our sense of safety feels threatened. There is no one standing apart from all this movement, pulling levers or pushing buttons--only patterns flowing through awareness itself.

The belief in a permanent self creates suffering because it separates us from life's endless transformations. But when we look closely enough, even this belief dissolves back into the stream where nothing remains fixed for more than an instant.

Example 3

Prompt: What do I do with anger when it arises?

When you notice the anger coming up, pause. Take one breath in through your nose if possible--this will help anchor you to this moment rather than carrying away into past or future scenarios. Now breathe out slowly while acknowledging what is present: 'I am aware of my anger.' This simple recognition creates space between feeling and acting on impulse. The anger may continue for a few seconds or minutes before settling naturally; that's okay too. You need not force anything different about how it manifests within you.

Training Summary

• Base model: HuggingFaceTB/SmolLM3-3B-Base
• Training format: V6E
• Teacher examples: 4,561
• Adapter recipe: QDoRA + rsLoRA
• Rank: 64
• Epochs: 1
• Completion-only loss: enabled

Run Metrics

• Run name: teacher-smollm3-v6e-tnh-l2-r64
• Training runtime: 42.3 minutes
• Final train loss: 1.6742
• Final eval loss: 1.6577
• Final eval token accuracy: 0.5669

Files

• TNH_Agent-SmolLM3-3B-v2-Q5_K_M.gguf (2.2 GB)

Honest Limits

• This is still a compact 3B model, so some broader contemplative phrasing can leak into non-teacher-specific prompts.
• The 5q smoke eval is encouraging, but it is not a full benchmark or final quality proof.

Public Naming

• Public repo: Meditation-Agent-TNH-SmolLM3-3B-v2-GGUF
• Internal model folder: TNH_Agent-SmolLM3-3B-v2