INDEX

Explanations

years from the 1980s

oai_token-act-pair · gpt-3.5-turbo

references to the 1980s

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

uden

-0.70

matically

-0.65

axy

-0.62

Hop

-0.61

hop

-0.59

actor

-0.59

 lawy

-0.59

 hops

-0.59

ailed

-0.57

arty

-0.56

POSITIVE LOGITS

rpm

0.81

peror

0.81

IRC

0.75

orpor

0.74

JV

0.73

ENSE

0.72

eenth

0.72

ption

0.72

ãĥĨãĤ£

0.72

ishly

0.71

Activations Density 0.084%

years from the 1980s

references to the 1980s

No Comments

No Known Activations