INDEX

Explanations

phrases related to government or military programs

oai_token-act-pair · gpt-3.5-turbo

references to various surveillance and military programs

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

pick

-0.75

ModLoader

-0.72

 dilig

-0.71

 humidity

-0.66

è¦

-0.65

angelo

-0.64

CLASSIFIED

-0.64

 Marketable

-0.63

poll

-0.63

picking

-0.63

POSITIVE LOGITS

mable

1.62

matically

1.38

matic

1.31

 Gutenberg

0.96

mers

0.91

eers

0.87

ional

0.80

heny

0.79

atical

0.79

 instituted

0.77

Activations Density 0.028%

phrases related to government or military programs

references to various surveillance and military programs

No Comments

No Known Activations