INDEX

Explanations

phrases related to criminal investigations and suspect activities

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

Its

-0.60

 gorilla

-0.59

 nutshell

-0.57

ucks

-0.54

 dude

-0.53

Its

-0.53

acus

-0.52

pires

-0.52

killer

-0.51

 rapist

-0.51

POSITIVE LOGITS

 respectively

1.23

 themselves

0.94

 apiece

0.92

 respective

0.91

 collectively

0.76

respective

0.76

their

0.73

 subsequently

0.73

 their

0.71

 individually

0.70

Activations Density 0.970%

phrases related to criminal investigations and suspect activities

No Comments

No Known Activations