INDEX

Explanations

years represented in dates

oai_token-act-pair · gpt-3.5-turbo

references to specific years and surveillance-related terms

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

ista

-0.88

eer

-0.86

isen

-0.85

apult

-0.84

aster

-0.82

zzi

-0.80

lain

-0.78

ingo

-0.77

inian

-0.74

ined

-0.74

POSITIVE LOGITS

rates

0.78

STATE

0.78

NESS

0.75

CEPT

0.74

FIL

0.73

 unfocusedRange

0.72

CLASSIFIED

0.68

ENC

0.66

Analysis

0.65

åŃ

0.65

Activations Density 0.061%

years represented in dates

references to specific years and surveillance-related terms

No Comments

No Known Activations