INDEX

Explanations

words related to daring or risky actions

oai_token-act-pair · gpt-3.5-turbo

references to stunts and pranks

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

oice

-0.72

soType

-0.69

inda

-0.68

ishops

-0.66

 gloom

-0.64

Ide

-0.63

 Sett

-0.63

oliberal

-0.62

iott

-0.62

tics

-0.61

POSITIVE LOGITS

hered

0.97

 stunts

0.91

 stunt

0.85

 performer

0.80

UAL

0.79

nel

0.77

strip

0.76

 performers

0.76

icular

0.74

 chore

0.74

Activations Density 0.062%

words related to daring or risky actions

references to stunts and pranks

No Comments

No Known Activations