INDEX

Explanations

words related to scams or deceitful behavior

oai_token-act-pair · gpt-3.5-turbo

references to swarming behavior or collective actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ICAL

-0.73

Reloaded

-0.71

OPLE

-0.70

 preferential

-0.70

pora

-0.70

_-

-0.70

 negligent

-0.67

ENSE

-0.66

NESS

-0.66

 flawed

-0.65

POSITIVE LOGITS

immers

1.06

inging

1.05

addle

1.04

arf

1.03

apon

1.03

anky

1.03

itched

1.00

imming

1.00

ashing

0.99

indle

0.99

Activations Density 0.010%

words related to scams or deceitful behavior

references to swarming behavior or collective actions

No Comments

No Known Activations