Research / Train models
GPU not connected
BROWSER-NATIVE LANGUAGE EXPERIMENTS

Same world.
Different word order.

Train two small language models from identical weights. Change the order of English clauses, then compare what the models actually write.

REAL BACKPROPAGATIONForward → loss → gradients → AdamW
causal transformer · f32
0 / 0no weights trained yet
word + punctuation tokens
non-padding tokens / second

01 Language profiles

same facts

Drag the clauses or use the arrows. Each block remains a grammatical English sentence. Only the order changes.

The first run checks numerical gradients and causal masking. Keep this tab open while training.

02 Live corpus preview

Preparing matched language…

Authored micro-stories, not human observations. The action patterns are taught through next-token text, not discovered moral rules.

03 Learning, not simulated progress

A · person-ledB · object-led
Cross-entropy in nats / predicted token. Lower is better.Dots: held-out canonical phrasing
Ready to initialise a matched pair
SAME PROMPT / DIFFERENT TRAINING HISTORIES

Ask. Compare. Inspect.

These are small, from-scratch language models—not pretrained chat assistants. Use the story examples first, then edit the task.

Input task

Relative candidate scores normalise length-averaged log-likelihoods over this list only. They are not calibrated probabilities of real-world actions.

Train a pair, then compare their generated language.No template answers will be substituted for the model’s output.
TOKEN-BY-TOKEN / CAUSALLY MASKED

Watch a prefix unfold.

Read the model’s next-word distribution at each input position, alongside actual attention weights. These are model measurements, not a visualisation of human thought.

Next-word probability

full vocabulary normalisation
The x-axis is the number of tokens read, not milliseconds. Seeing a person earlier is not by itself evidence of a learned person-first preference.

Last-layer attention

Heads averaged. Rows query earlier columns. The empty upper triangle is the causal mask. Hover for token pairs.

No trace yet

Prefix readout

Token, highest-probability next word and distribution entropy.

AUDIT THE EXPERIMENT

Nothing behind the curtain.

Actual weights, gradients, source code and provenance. This is a controlled computational language experiment.

What is controlled

Initialisation
Byte-identical weight arrays for both arms. No dropout.
Experiences
Same event IDs, same minibatch order, same continuation text, same number of non-padding tokens.
Language treatment
Natural English clause order. Clauses are not omitted or rewritten. Word counts match exactly.
Held-out data
Exact story combinations are disjoint. Templates and vocabulary intentionally overlap.
Evaluation
Common prompts are contrasted with native-order prompts. Shared canonical training exposure is configurable.
Interpretation
One pair is exploratory. More seeds and same-grammar controls are needed before attributing a systematic effect to the treatment.

What this does not claim

No proof about human brains, national cultures, political beliefs or moral behaviour. No pretrained world knowledge. Word order here means the order of complete English clauses—not a fully constructed SOV or VSO language.

Generated explanations are text predictions, not access to an internal reasoning process. Attention weights are not a causal explanation. Out-of-vocabulary words are explicitly shown rather than silently understood.

Training examples contain authored action consequences. Their equal use across arms is controlled; they are not value-free examples of the real world.

Numerical validation

Identical initialisation • future-token leakage • finite-difference gradients • actual loss reduction • checkpoint and optimizer restoration.

Not run in this browser.

Kernel inspector

CUDA source

WGSL reference

20 CUDA kernels implement embedding, tiled matmul, RMSNorm, causal attention, SiLU, cross-entropy, all gradients, clipping and AdamW. No inference API or hidden CPU training.

Session log No activity