INDEX

Explanations

questions and expressions of uncertainty

oai_token-act-pair · gpt-4o-mini Triggered by @bot

asking a question

np_acts-logits-general · gemini-2.5-flash-lite

This neuron detects message/turn boundaries and large blank/formatting breaks (i.e., paragraph or post separators) in multi-turn/forum-style text.

oai_token-act-pair · gpt-5-mini Triggered by @vetterc0

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 inginkan

-0.36

in

-0.33

tritts

-0.32

 into

-0.31

but

-0.30

 keinem

-0.29

 apapun

-0.29

 single

-0.29

 død

-0.29

 tương

-0.28

POSITIVE LOGITS

tvguidetime

0.75

 nakalista

0.74

 accountId

0.64

 ſoll

0.59

 dieſen

0.57

ðsíða

0.57

 queſta

0.56

 müſſen

0.56

 مرئيه

0.56

EndProject

0.56

Activations Density 0.035%

questions and expressions of uncertainty

asking a question

This neuron detects message/turn boundaries and large blank/formatting breaks (i.e., paragraph or post separators) in multi-turn/forum-style text.

No Comments

No Known Activations