INDEX

Explanations

terms related to observing or being present for events

oai_token-act-pair · gpt-3.5-turbo

instances of the word "witnessed."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

afort

-0.92

bit

-0.72

adies

-0.72

imb

-0.71

icrobial

-0.70

cise

-0.69

separ

-0.69

neg

-0.68

processing

-0.68

hate

-0.68

POSITIVE LOGITS

ewitness

1.04

ĸļ

1.02

 witnessed

0.99

 firsthand

0.93

 witnessing

0.84

-+-+

0.81

wcsstore

0.80

¿½

0.80

 witness

0.78

 witnesses

0.76

Activations Density 0.006%

terms related to observing or being present for events

instances of the word "witnessed."

No Comments

No Known Activations