INDEX

Explanations

violent or destructive actions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "broke" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

ulp

-0.74

 Colonial

-0.61

metics

-0.60

ogical

-0.60

idency

-0.60

apixel

-0.60

alogy

-0.59

etheless

-0.59

imental

-0.59

alty

-0.59

POSITIVE LOGITS

neck

0.95

 broke

0.94

breaks

0.91

break

0.83

 breaks

0.83

 break

0.82

breakers

0.79

bre

0.75

rael

0.74

chain

0.71

Activations Density 0.021%

violent or destructive actions

instances of the word "broke" in various contexts

No Comments

No Known Activations