INDEX

Explanations

words related to specific locations, potentially with references to events or people associated with them

oai_token-act-pair · gpt-3.5-turbo

references to proper nouns or entities, particularly names and titles

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

theless

-0.65

REDACTED

-0.63

 Sina

-0.62

ablishment

-0.61

 props

-0.60

 GOODMAN

-0.59

 cops

-0.59

 kids

-0.58

 advertisers

-0.58

IUM

-0.57

POSITIVE LOGITS

utsch

0.97

ymes

0.91

astery

0.89

eworks

0.86

actor

0.85

ule

0.84

iasco

0.80

utor

0.79

oub

0.78

iner

0.77

Activations Density 0.272%

words related to specific locations, potentially with references to events or people associated with them

references to proper nouns or entities, particularly names and titles

No Comments

No Known Activations