INDEX

Explanations

phrases related to potential negative events or disasters

oai_token-act-pair · gpt-3.5-turbo

contexts related to events or emergencies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

kefeller

-0.90

educated

-0.70

pees

-0.70

sort

-0.66

 srfAttach

-0.66

ptives

-0.65

ometry

-0.65

ilib

-0.64

Pub

-0.64

Dub

-0.64

POSITIVE LOGITS

 malf

1.14

 emergencies

1.05

 misfortune

1.04

 malfunction

1.02

 unforeseen

1.01

 unexpectedly

1.00

 stray

0.99

 disagreement

0.97

 misconduct

0.96

 mish

0.93

Activations Density 0.785%

phrases related to potential negative events or disasters

contexts related to events or emergencies

No Comments

No Known Activations