INDEX

Explanations

the word "and" in various contexts

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

phrases with and

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

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

 otomatig

-0.62

 itſelf

-0.57

 myſelf

-0.55

writeFieldEnd

-0.53

 ſtate

-0.52

bibfield

-0.52

onAttach

-0.52

 ſche

-0.52

HostException

-0.50

 ainfi

-0.50

POSITIVE LOGITS

 With

0.49

 with

0.47

 עם

0.46

with

0.45

 ample

0.44

With

0.44

 fresh

0.43

 WITH

0.43

WithMany

0.41

Activations Density 0.031%

the word "and" in various contexts

phrases with and

No Comments

No Known Activations