INDEX

Explanations

the word "for" indicating responsibility or attribution

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 0-gemmascope-mlp-65k

Configuration

google/gemma-scope-2b-pt-mlp/layer_0/width_65k/average_l0_72

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.0.hook_mlp_out

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 TÉCN

-0.73

disfraz

-0.68

Efq

-0.66

 myſelf

-0.66

ſelves

-0.64

^(@)

-0.60

 itſelf

-0.60

↵

-0.60

 Jefus

-0.60

inerja

-0.59

POSITIVE LOGITS

setVerticalGroup

0.61

WTS

0.59

Cu

0.56

 biele

0.55

long

0.55

InSection

0.54

uck

0.54

BASELINE

0.53

ود

0.52

mis

0.52

Activations Density 0.062%

the word "for" indicating responsibility or attribution

No Comments

No Known Activations