INDEX

Explanations

references to reported or alleged actions

oai_token-act-pair · gpt-3.5-turbo

phrases indicating reports or allegations about events or actions

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Appearances

-0.81

bern

-0.81

uit

-0.78

 surely

-0.78

vation

-0.75

ERAL

-0.74

cair

-0.74

etime

-0.71

egg

-0.69

usually

-0.69

POSITIVE LOGITS

 violated

0.82

 overheard

0.78

 threatened

0.76

 instructed

0.74

 smuggled

0.74

 coerced

0.73

 involved

0.73

 sabot

0.71

 mist

0.71

 stole

0.71

Activations Density 0.054%

references to reported or alleged actions

phrases indicating reports or allegations about events or actions

No Comments

No Known Activations