INDEX

Explanations

terms related to abstract concepts or fields of study, particularly those related to intellectual realms

oai_token-act-pair · gpt-3.5-turbo

mentions of various "realms" or domains of experience

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ãģį

-0.66

HOME

-0.63

 Clover

-0.62

bor

-0.62

orah

-0.61

 Soda

-0.61

INST

-0.60

{\

-0.59

berman

-0.59

 Rate

-0.59

POSITIVE LOGITS

naire

0.93

rums

0.84

 realms

0.82

uin

0.82

 collide

0.80

mares

0.76

 realm

0.76

istry

0.75

osaurs

0.74

icular

0.74

Activations Density 0.028%

terms related to abstract concepts or fields of study, particularly those related to intellectual realms

mentions of various "realms" or domains of experience

No Comments

No Known Activations