INDEX

Explanations

instances where something is declared as permissible or free from blame or suspicion

oai_token-act-pair · gpt-3.5-turbo

instances of the word "cleared" in various contexts

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

ENTS

-0.75

ENT

-0.75

egal

-0.74

ENC

-0.70

INA

-0.67

lav

-0.64

ADD

-0.63

EG

-0.61

ingo

-0.61

DOM

-0.61

POSITIVE LOGITS

 clears

1.01

 cleared

0.93

 waivers

0.89

0.79

houses

0.78

 paperwork

0.76

ĵĺ

0.74

 clearance

0.74

sheets

0.74

 Clear

0.73

Activations Density 0.015%

instances where something is declared as permissible or free from blame or suspicion

instances of the word "cleared" in various contexts

No Comments

No Known Activations