eisv-lumen-teacher

EISV-Lumen Teacher — LoRA on Qwen3-4B

A LoRA fine-tune of Qwen/Qwen3-4B that maps EISV trajectory shapes to primitive expression tokens. Given a sequence of EISV dimension values and a classified trajectory shape, the model generates a structured expression — a short sequence of emotion-like tokens that a downstream embodied system (Lumen, running on a Raspberry Pi 4) uses to drive LED color, screen drawing, and inner-voice narration.

This is the teacher model in a teacher-student pipeline. Its outputs are distilled into a lightweight RandomForest student that runs on-device.

Model Details

What is EISV?

EISV is a four-coordinate state estimate. In the real-data portion of this model's training pipeline, the coordinates are derived from Lumen's physical sensors and computational telemetry. The training set also includes synthetic trajectories; sensor provenance does not apply to those examples.

The companion dataset card documents the Lumen observation-layer mapping:

The coordinate names and expression tokens are modeling conventions, not measurements of subjective experience. Expression coherence below is a token-to-shape fit score, distinct from UNITARES governance coherence; it does not validate safety or reasoning quality.

Training provenance and sampling cadence

The v6 training configuration describes a blend of 1,803 real-derived and 1,797 synthetic examples before splitting. The retained local files contain 2,880 training, 360 validation, and 360 test examples. The linked Hub dataset is a larger trajectory-window corpus, not the 2,880-example training split itself.

The pinned companion dataset card reports typical sensor-driven intervals of approximately 2 seconds, with actual timestamps in each window. The earlier 30-second claim in this model card was not tied to a documented sampling or downsampling stage and has been removed. This dataset-card revision documents the coordinate mapping and corpus; it is not a verified pin of the teacher's original training input. The retained formatted training files contain text, shape, pattern, and messages but no source timestamps or Hub revision, so they do not establish the original training-window cadence or exact upstream dataset cut.

Trajectory Shapes

A trajectory is a time-window of EISV values. The shape classifier identifies 9 patterns. The frequencies below are approximate historical corpus summaries, not counts for the current Hub dataset or the balanced training split. Shape names describe coordinate patterns, not independently measured psychological states:

The distribution is heavily skewed toward settled_presence and convergence in the historical corpus; this distribution alone does not establish device health or agent capability. The training pipeline uses approximately 50/50 real+synthetic blending to ensure the model sees enough rare shapes.

Expression Vocabulary

The model generates structured expressions using 15 primitive tokens and 5 patterns.

Tokens (15)

~warmth~ ~curiosity~ ~resonance~ ~stillness~ ~boundary~ ~reaching~ ~reflection~ ~ripple~ ~deep_listening~ ~emergence~ ~questioning~ ~holding~ ~releasing~ ~threshold~ ~return~

Patterns (5)

Evaluation Results

Evaluated on 500 real trajectories (no synthetic data in the test set):

Coherence measures how well the generated expression fits the trajectory shape (semantic alignment between tokens and shape meaning). Valid rate measures whether outputs parse as legal expressions. Pattern accuracy measures exact match of the structural pattern — the relatively low score is expected since multiple patterns can be coherent for a given shape.

Per-Shape Coherence

Usage

from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

# Load base model
base_model = AutoModelForCausalLM.from_pretrained(
    "Qwen/Qwen3-4B",
    torch_dtype="auto",
    device_map="auto",
)
tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen3-4B")

# Apply LoRA adapter
model = PeftModel.from_pretrained(base_model, "hikewa/eisv-lumen-teacher")

# Format input prompt
prompt = """You are an EISV expression generator. Given trajectory dynamics, produce a primitive expression.

Trajectory shape: settled_presence
EISV values: E=0.72, I=0.85, S=0.15, V=0.08
Dominant dimension: I (Information Integrity)
Trend: stable

Expression:"""

inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(
    **inputs,
    max_new_tokens=32,
    temperature=0.7,
    do_sample=True,
)
result = tokenizer.decode(outputs[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True)
print(result)
# Example output: ~stillness~ ~resonance~

Related

• Dataset: hikewa/unitares-eisv-trajectories — companion trajectory-window corpus; see the training-provenance limits above
• Student model: hikewa/eisv-lumen-student — distilled RandomForest that runs on Raspberry Pi 4
• Demo: hikewa/eisv-lumen-explorer — interactive Space to explore trajectories and expressions

Citation

@misc{eisv-lumen-teacher-2026,
  title={EISV-Lumen Teacher: LoRA Fine-Tune for Trajectory-to-Expression Generation},
  author={Wang, Kenny},
  year={2026},
  url={https://huggingface.co/hikewa/eisv-lumen-teacher},
  note={LoRA adapter on Qwen/Qwen3-4B mapping EISV trajectory shapes to primitive expression tokens}
}