INDEX

Explanations

This neuron seems to activate on profanity as well as elements of programming languages, math formulas, measurements, and questions about probability.

oai_token-act-pair · gemini-2.0-flash

New Auto-Interp

Configuration

google/gemma-scope-2b-pt-transcoders/layer_7/width_16k/average_l0_70

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.7.ln2.hook_normalized

Architecture

jumprelu_transcoder

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 consigo

-0.46

堂

-0.45

Ar

-0.44

Mal

-0.42

-0.41

Mar

-0.40

案

-0.40

ran

-0.40

POSITIVE LOGITS

 myſelf

1.02

Efq

1.00

хьтан

0.96

 itſelf

0.96

 whoſe

0.93

writeFieldEnd

0.92

 ſta

0.89

 חיצוניים

0.89

 ſeveral

0.89

 diſt

0.88

Activations Density 2.833%

This neuron seems to activate on profanity as well as elements of programming languages, math formulas, measurements, and questions about probability.

No Comments

No Known Activations